[Summary view] [Print] [Text view]

   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 <[email protected]>
  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      /** An 8-bit (1-byte) unsigned integer. */
  34      const BYTE = 1;
  35  
  36      /** An 8-bit byte containing one 7-bit ASCII code.
  37       *  The final byte is terminated with NULL.
  38       */
  39      const ASCII = 2;
  40  
  41      /** A 16-bit (2-byte) unsigned integer. */
  42      const SHORT = 3;
  43  
  44      /** A 32-bit (4-byte) unsigned integer. */
  45      const LONG = 4;
  46  
  47      /** Two LONGs. The first LONG is the numerator and the second LONG expresses
  48       *  the denominator
  49       */
  50      const RATIONAL = 5;
  51  
  52      /** A 16-bit (2-byte) or 32-bit (4-byte) unsigned integer. */
  53      const SHORT_OR_LONG = 6;
  54  
  55      /** An 8-bit byte that can take any value depending on the field definition */
  56      const UNDEFINED = 7;
  57  
  58      /** A 32-bit (4-byte) signed integer (2's complement notation), */
  59      const SLONG = 9;
  60  
  61      /** Two SLONGs. The first SLONG is the numerator and the second SLONG is
  62       *  the denominator.
  63       */
  64      const SRATIONAL = 10;
  65  
  66      /** A fake value for things we don't want or don't support. */
  67      const IGNORE = -1;
  68  
  69      /** @var array Exif tags grouped by category, the tagname itself is the key
  70       *    and the type is the value, in the case of more than one possible value
  71       *    type they are separated by commas.
  72       */
  73      private $mExifTags;
  74  
  75      /** @var array The raw Exif data returned by exif_read_data() */
  76      private $mRawExifData;
  77  
  78      /** @var array A Filtered version of $mRawExifData that has been pruned
  79       *    of invalid tags and tags that contain content they shouldn't contain
  80       *    according to the Exif specification
  81       */
  82      private $mFilteredExifData;
  83  
  84      /** @var string The file being processed */
  85      private $file;
  86  
  87      /** @var string The basename of the file being processed */
  88      private $basename;
  89  
  90      /** @var string The private log to log to, e.g. 'exif' */
  91      private $log = false;
  92  
  93      /** @var string The byte order of the file. Needed because php's extension
  94       *    doesn't fully process some obscure props.
  95       */
  96      private $byteOrder;
  97  
  98      /**
  99       * Constructor
 100       *
 101       * @param string $file Filename.
 102       * @param string $byteOrder Type of byte ordering either 'BE' (Big Endian)
 103       *   or 'LE' (Little Endian). Default ''.
 104       * @throws MWException
 105       * @todo FIXME: The following are broke:
 106       *   SubjectArea. Need to test the more obscure tags.
 107       *   DigitalZoomRatio = 0/0 is rejected. need to determine if that's valid.
 108       *   Possibly should treat 0/0 = 0. need to read exif spec on that.
 109       */
 110  	function __construct( $file, $byteOrder = '' ) {
 111          /**
 112           * Page numbers here refer to pages in the Exif 2.2 standard
 113           *
 114           * Note, Exif::UNDEFINED is treated as a string, not as an array of bytes
 115           * so don't put a count parameter for any UNDEFINED values.
 116           *
 117           * @link http://exif.org/Exif2-2.PDF The Exif 2.2 specification
 118           */
 119          $this->mExifTags = array(
 120              # TIFF Rev. 6.0 Attribute Information (p22)
 121              'IFD0' => array(
 122                  # Tags relating to image structure
 123                  'ImageWidth' => Exif::SHORT_OR_LONG, # Image width
 124                  'ImageLength' => Exif::SHORT_OR_LONG, # Image height
 125                  'BitsPerSample' => array( Exif::SHORT, 3 ), # Number of bits per component
 126                  # "When a primary image is JPEG compressed, this designation is not"
 127                  # "necessary and is omitted." (p23)
 128                  'Compression' => Exif::SHORT, # Compression scheme #p23
 129                  'PhotometricInterpretation' => Exif::SHORT, # Pixel composition #p23
 130                  'Orientation' => Exif::SHORT, # Orientation of image #p24
 131                  'SamplesPerPixel' => Exif::SHORT, # Number of components
 132                  'PlanarConfiguration' => Exif::SHORT, # Image data arrangement #p24
 133                  'YCbCrSubSampling' => array( Exif::SHORT, 2 ), # Subsampling ratio of Y to C #p24
 134                  'YCbCrPositioning' => Exif::SHORT, # Y and C positioning #p24-25
 135                  'XResolution' => Exif::RATIONAL, # Image resolution in width direction
 136                  'YResolution' => Exif::RATIONAL, # Image resolution in height direction
 137                  'ResolutionUnit' => Exif::SHORT, # Unit of X and Y resolution #(p26)
 138  
 139                  # Tags relating to recording offset
 140                  'StripOffsets' => Exif::SHORT_OR_LONG, # Image data location
 141                  'RowsPerStrip' => Exif::SHORT_OR_LONG, # Number of rows per strip
 142                  'StripByteCounts' => Exif::SHORT_OR_LONG, # Bytes per compressed strip
 143                  'JPEGInterchangeFormat' => Exif::SHORT_OR_LONG, # Offset to JPEG SOI
 144                  'JPEGInterchangeFormatLength' => Exif::SHORT_OR_LONG, # Bytes of JPEG data
 145  
 146                  # Tags relating to image data characteristics
 147                  'TransferFunction' => Exif::IGNORE, # Transfer function
 148                  'WhitePoint' => array( Exif::RATIONAL, 2 ), # White point chromaticity
 149                  'PrimaryChromaticities' => array( Exif::RATIONAL, 6 ), # Chromaticities of primarities
 150                  # Color space transformation matrix coefficients #p27
 151                  'YCbCrCoefficients' => array( Exif::RATIONAL, 3 ),
 152                  'ReferenceBlackWhite' => array( Exif::RATIONAL, 6 ), # Pair of black and white reference values
 153  
 154                  # Other tags
 155                  'DateTime' => Exif::ASCII, # File change date and time
 156                  'ImageDescription' => Exif::ASCII, # Image title
 157                  'Make' => Exif::ASCII, # Image input equipment manufacturer
 158                  'Model' => Exif::ASCII, # Image input equipment model
 159                  'Software' => Exif::ASCII, # Software used
 160                  'Artist' => Exif::ASCII, # Person who created the image
 161                  'Copyright' => Exif::ASCII, # Copyright holder
 162              ),
 163  
 164              # Exif IFD Attribute Information (p30-31)
 165              'EXIF' => array(
 166                  # @todo NOTE: Nonexistence of this field is taken to mean nonconformance
 167                  # to the Exif 2.1 AND 2.2 standards
 168                  'ExifVersion' => Exif::UNDEFINED, # Exif version
 169                  'FlashPixVersion' => Exif::UNDEFINED, # Supported Flashpix version #p32
 170  
 171                  # Tags relating to Image Data Characteristics
 172                  'ColorSpace' => Exif::SHORT, # Color space information #p32
 173  
 174                  # Tags relating to image configuration
 175                  'ComponentsConfiguration' => Exif::UNDEFINED, # Meaning of each component #p33
 176                  'CompressedBitsPerPixel' => Exif::RATIONAL, # Image compression mode
 177                  'PixelYDimension' => Exif::SHORT_OR_LONG, # Valid image width
 178                  'PixelXDimension' => Exif::SHORT_OR_LONG, # Valid image height
 179  
 180                  # Tags relating to related user information
 181                  'MakerNote' => Exif::IGNORE, # Manufacturer notes
 182                  'UserComment' => Exif::UNDEFINED, # User comments #p34
 183  
 184                  # Tags relating to related file information
 185                  'RelatedSoundFile' => Exif::ASCII, # Related audio file
 186  
 187                  # Tags relating to date and time
 188                  'DateTimeOriginal' => Exif::ASCII, # Date and time of original data generation #p36
 189                  'DateTimeDigitized' => Exif::ASCII, # Date and time of original data generation
 190                  'SubSecTime' => Exif::ASCII, # DateTime subseconds
 191                  'SubSecTimeOriginal' => Exif::ASCII, # DateTimeOriginal subseconds
 192                  'SubSecTimeDigitized' => Exif::ASCII, # DateTimeDigitized subseconds
 193  
 194                  # Tags relating to picture-taking conditions (p31)
 195                  'ExposureTime' => Exif::RATIONAL, # Exposure time
 196                  'FNumber' => Exif::RATIONAL, # F Number
 197                  'ExposureProgram' => Exif::SHORT, # Exposure Program #p38
 198                  'SpectralSensitivity' => Exif::ASCII, # Spectral sensitivity
 199                  'ISOSpeedRatings' => Exif::SHORT, # ISO speed rating
 200                  'OECF' => Exif::IGNORE,
 201                  # Optoelectronic conversion factor. Note: We don't have support for this atm.
 202                  'ShutterSpeedValue' => Exif::SRATIONAL, # Shutter speed
 203                  'ApertureValue' => Exif::RATIONAL, # Aperture
 204                  'BrightnessValue' => Exif::SRATIONAL, # Brightness
 205                  'ExposureBiasValue' => Exif::SRATIONAL, # Exposure bias
 206                  'MaxApertureValue' => Exif::RATIONAL, # Maximum land aperture
 207                  'SubjectDistance' => Exif::RATIONAL, # Subject distance
 208                  'MeteringMode' => Exif::SHORT, # Metering mode #p40
 209                  'LightSource' => Exif::SHORT, # Light source #p40-41
 210                  'Flash' => Exif::SHORT, # Flash #p41-42
 211                  'FocalLength' => Exif::RATIONAL, # Lens focal length
 212                  'SubjectArea' => array( Exif::SHORT, 4 ), # Subject area
 213                  'FlashEnergy' => Exif::RATIONAL, # Flash energy
 214                  'SpatialFrequencyResponse' => Exif::IGNORE, # Spatial frequency response. Not supported atm.
 215                  'FocalPlaneXResolution' => Exif::RATIONAL, # Focal plane X resolution
 216                  'FocalPlaneYResolution' => Exif::RATIONAL, # Focal plane Y resolution
 217                  'FocalPlaneResolutionUnit' => Exif::SHORT, # Focal plane resolution unit #p46
 218                  'SubjectLocation' => array( Exif::SHORT, 2 ), # Subject location
 219                  'ExposureIndex' => Exif::RATIONAL, # Exposure index
 220                  'SensingMethod' => Exif::SHORT, # Sensing method #p46
 221                  'FileSource' => Exif::UNDEFINED, # File source #p47
 222                  'SceneType' => Exif::UNDEFINED, # Scene type #p47
 223                  'CFAPattern' => Exif::IGNORE, # CFA pattern. not supported atm.
 224                  'CustomRendered' => Exif::SHORT, # Custom image processing #p48
 225                  'ExposureMode' => Exif::SHORT, # Exposure mode #p48
 226                  'WhiteBalance' => Exif::SHORT, # White Balance #p49
 227                  'DigitalZoomRatio' => Exif::RATIONAL, # Digital zoom ration
 228                  'FocalLengthIn35mmFilm' => Exif::SHORT, # Focal length in 35 mm film
 229                  'SceneCaptureType' => Exif::SHORT, # Scene capture type #p49
 230                  'GainControl' => Exif::SHORT, # Scene control #p49-50
 231                  'Contrast' => Exif::SHORT, # Contrast #p50
 232                  'Saturation' => Exif::SHORT, # Saturation #p50
 233                  'Sharpness' => Exif::SHORT, # Sharpness #p50
 234                  'DeviceSettingDescription' => Exif::IGNORE,
 235                  # Device settings description. This could maybe be supported. Need to find an
 236                  # example file that uses this to see if it has stuff of interest in it.
 237                  'SubjectDistanceRange' => Exif::SHORT, # Subject distance range #p51
 238  
 239                  'ImageUniqueID' => Exif::ASCII, # Unique image ID
 240              ),
 241  
 242              # GPS Attribute Information (p52)
 243              'GPS' => array(
 244                  'GPSVersion' => Exif::UNDEFINED,
 245                  # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 246                  # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 247                  'GPSLatitudeRef' => Exif::ASCII, # North or South Latitude #p52-53
 248                  'GPSLatitude' => array( Exif::RATIONAL, 3 ), # Latitude
 249                  'GPSLongitudeRef' => Exif::ASCII, # East or West Longitude #p53
 250                  'GPSLongitude' => array( Exif::RATIONAL, 3 ), # Longitude
 251                  'GPSAltitudeRef' => Exif::UNDEFINED,
 252                  # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 253                  # but php seems to disagree.
 254                  'GPSAltitude' => Exif::RATIONAL, # Altitude
 255                  'GPSTimeStamp' => array( Exif::RATIONAL, 3 ), # GPS time (atomic clock)
 256                  'GPSSatellites' => Exif::ASCII, # Satellites used for measurement
 257                  'GPSStatus' => Exif::ASCII, # Receiver status #p54
 258                  'GPSMeasureMode' => Exif::ASCII, # Measurement mode #p54-55
 259                  'GPSDOP' => Exif::RATIONAL, # Measurement precision
 260                  'GPSSpeedRef' => Exif::ASCII, # Speed unit #p55
 261                  'GPSSpeed' => Exif::RATIONAL, # Speed of GPS receiver
 262                  'GPSTrackRef' => Exif::ASCII, # Reference for direction of movement #p55
 263                  'GPSTrack' => Exif::RATIONAL, # Direction of movement
 264                  'GPSImgDirectionRef' => Exif::ASCII, # Reference for direction of image #p56
 265                  'GPSImgDirection' => Exif::RATIONAL, # Direction of image
 266                  'GPSMapDatum' => Exif::ASCII, # Geodetic survey data used
 267                  'GPSDestLatitudeRef' => Exif::ASCII, # Reference for latitude of destination #p56
 268                  'GPSDestLatitude' => array( Exif::RATIONAL, 3 ), # Latitude destination
 269                  'GPSDestLongitudeRef' => Exif::ASCII, # Reference for longitude of destination #p57
 270                  'GPSDestLongitude' => array( Exif::RATIONAL, 3 ), # Longitude of destination
 271                  'GPSDestBearingRef' => Exif::ASCII, # Reference for bearing of destination #p57
 272                  'GPSDestBearing' => Exif::RATIONAL, # Bearing of destination
 273                  'GPSDestDistanceRef' => Exif::ASCII, # Reference for distance to destination #p57-58
 274                  'GPSDestDistance' => Exif::RATIONAL, # Distance to destination
 275                  'GPSProcessingMethod' => Exif::UNDEFINED, # Name of GPS processing method
 276                  'GPSAreaInformation' => Exif::UNDEFINED, # Name of GPS area
 277                  'GPSDateStamp' => Exif::ASCII, # GPS date
 278                  'GPSDifferential' => Exif::SHORT, # GPS differential correction
 279              ),
 280          );
 281  
 282          $this->file = $file;
 283          $this->basename = wfBaseName( $this->file );
 284          if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 285              $this->byteOrder = $byteOrder;
 286          } else {
 287              // Only give a warning for b/c, since originally we didn't
 288              // require this. The number of things affected by this is
 289              // rather small.
 290              wfWarn( 'Exif class did not have byte order specified. ' .
 291                  'Some properties may be decoded incorrectly.' );
 292              $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 293          }
 294  
 295          $this->debugFile( $this->basename, __FUNCTION__, true );
 296          if ( function_exists( 'exif_read_data' ) ) {
 297              wfSuppressWarnings();
 298              $data = exif_read_data( $this->file, 0, true );
 299              wfRestoreWarnings();
 300          } else {
 301              throw new MWException( "Internal error: exif_read_data not present. " .
 302                  "\$wgShowEXIF may be incorrectly set or not checked by an extension." );
 303          }
 304          /**
 305           * exif_read_data() will return false on invalid input, such as
 306           * when somebody uploads a file called something.jpeg
 307           * containing random gibberish.
 308           */
 309          $this->mRawExifData = $data ?: array();
 310          $this->makeFilteredData();
 311          $this->collapseData();
 312          $this->debugFile( __FUNCTION__, false );
 313      }
 314  
 315      /**
 316       * Make $this->mFilteredExifData
 317       */
 318  	function makeFilteredData() {
 319          $this->mFilteredExifData = array();
 320  
 321          foreach ( array_keys( $this->mRawExifData ) as $section ) {
 322              if ( !array_key_exists( $section, $this->mExifTags ) ) {
 323                  $this->debug( $section, __FUNCTION__, "'$section' is not a valid Exif section" );
 324                  continue;
 325              }
 326  
 327              foreach ( array_keys( $this->mRawExifData[$section] ) as $tag ) {
 328                  if ( !array_key_exists( $tag, $this->mExifTags[$section] ) ) {
 329                      $this->debug( $tag, __FUNCTION__, "'$tag' is not a valid tag in '$section'" );
 330                      continue;
 331                  }
 332  
 333                  $this->mFilteredExifData[$tag] = $this->mRawExifData[$section][$tag];
 334                  // This is ok, as the tags in the different sections do not conflict.
 335                  // except in computed and thumbnail section, which we don't use.
 336  
 337                  $value = $this->mRawExifData[$section][$tag];
 338                  if ( !$this->validate( $section, $tag, $value ) ) {
 339                      $this->debug( $value, __FUNCTION__, "'$tag' contained invalid data" );
 340                      unset( $this->mFilteredExifData[$tag] );
 341                  }
 342              }
 343          }
 344      }
 345  
 346      /**
 347       * Collapse some fields together.
 348       * This converts some fields from exif form, to a more friendly form.
 349       * For example GPS latitude to a single number.
 350       *
 351       * The rationale behind this is that we're storing data, not presenting to the user
 352       * For example a longitude is a single number describing how far away you are from
 353       * the prime meridian. Well it might be nice to split it up into minutes and seconds
 354       * for the user, it doesn't really make sense to split a single number into 4 parts
 355       * for storage. (degrees, minutes, second, direction vs single floating point number).
 356       *
 357       * Other things this might do (not really sure if they make sense or not):
 358       * Dates -> mediawiki date format.
 359       * convert values that can be in different units to be in one standardized unit.
 360       *
 361       * As an alternative approach, some of this could be done in the validate phase
 362       * if we make up our own types like Exif::DATE.
 363       */
 364  	function collapseData() {
 365  
 366          $this->exifGPStoNumber( 'GPSLatitude' );
 367          $this->exifGPStoNumber( 'GPSDestLatitude' );
 368          $this->exifGPStoNumber( 'GPSLongitude' );
 369          $this->exifGPStoNumber( 'GPSDestLongitude' );
 370  
 371          if ( isset( $this->mFilteredExifData['GPSAltitude'] )
 372              && isset( $this->mFilteredExifData['GPSAltitudeRef'] )
 373          ) {
 374              // We know altitude data is a <num>/<denom> from the validation
 375              // functions ran earlier. But multiplying such a string by -1
 376              // doesn't work well, so convert.
 377              list( $num, $denom ) = explode( '/', $this->mFilteredExifData['GPSAltitude'] );
 378              $this->mFilteredExifData['GPSAltitude'] = $num / $denom;
 379  
 380              if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 381                  $this->mFilteredExifData['GPSAltitude'] *= -1;
 382              }
 383              unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 384          }
 385  
 386          $this->exifPropToOrd( 'FileSource' );
 387          $this->exifPropToOrd( 'SceneType' );
 388  
 389          $this->charCodeString( 'UserComment' );
 390          $this->charCodeString( 'GPSProcessingMethod' );
 391          $this->charCodeString( 'GPSAreaInformation' );
 392  
 393          //ComponentsConfiguration should really be an array instead of a string...
 394          //This turns a string of binary numbers into an array of numbers.
 395  
 396          if ( isset( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 397              $val = $this->mFilteredExifData['ComponentsConfiguration'];
 398              $ccVals = array();
 399  
 400              $strLen = strlen( $val );
 401              for ( $i = 0; $i < $strLen; $i++ ) {
 402                  $ccVals[$i] = ord( substr( $val, $i, 1 ) );
 403              }
 404              $ccVals['_type'] = 'ol'; //this is for formatting later.
 405              $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 406          }
 407  
 408          //GPSVersion(ID) is treated as the wrong type by php exif support.
 409          //Go through each byte turning it into a version string.
 410          //For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 411  
 412          //Also change exif tag name from GPSVersion (what php exif thinks it is)
 413          //to GPSVersionID (what the exif standard thinks it is).
 414  
 415          if ( isset( $this->mFilteredExifData['GPSVersion'] ) ) {
 416              $val = $this->mFilteredExifData['GPSVersion'];
 417              $newVal = '';
 418  
 419              $strLen = strlen( $val );
 420              for ( $i = 0; $i < $strLen; $i++ ) {
 421                  if ( $i !== 0 ) {
 422                      $newVal .= '.';
 423                  }
 424                  $newVal .= ord( substr( $val, $i, 1 ) );
 425              }
 426  
 427              if ( $this->byteOrder === 'LE' ) {
 428                  // Need to reverse the string
 429                  $newVal2 = '';
 430                  for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 431                      $newVal2 .= substr( $newVal, $i, 1 );
 432                  }
 433                  $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 434              } else {
 435                  $this->mFilteredExifData['GPSVersionID'] = $newVal;
 436              }
 437              unset( $this->mFilteredExifData['GPSVersion'] );
 438          }
 439      }
 440  
 441      /**
 442       * Do userComment tags and similar. See pg. 34 of exif standard.
 443       * basically first 8 bytes is charset, rest is value.
 444       * This has not been tested on any shift-JIS strings.
 445       * @param string $prop Prop name
 446       */
 447  	private function charCodeString( $prop ) {
 448          if ( isset( $this->mFilteredExifData[$prop] ) ) {
 449  
 450              if ( strlen( $this->mFilteredExifData[$prop] ) <= 8 ) {
 451                  //invalid. Must be at least 9 bytes long.
 452  
 453                  $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, false );
 454                  unset( $this->mFilteredExifData[$prop] );
 455  
 456                  return;
 457              }
 458              $charCode = substr( $this->mFilteredExifData[$prop], 0, 8 );
 459              $val = substr( $this->mFilteredExifData[$prop], 8 );
 460  
 461              switch ( $charCode ) {
 462                  case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 463                      //JIS
 464                      $charset = "Shift-JIS";
 465                      break;
 466                  case "UNICODE\x00":
 467                      $charset = "UTF-16" . $this->byteOrder;
 468                      break;
 469                  default: //ascii or undefined.
 470                      $charset = "";
 471                      break;
 472              }
 473              if ( $charset ) {
 474                  wfSuppressWarnings();
 475                  $val = iconv( $charset, 'UTF-8//IGNORE', $val );
 476                  wfRestoreWarnings();
 477              } else {
 478                  // if valid utf-8, assume that, otherwise assume windows-1252
 479                  $valCopy = $val;
 480                  UtfNormal::quickIsNFCVerify( $valCopy ); //validates $valCopy.
 481                  if ( $valCopy !== $val ) {
 482                      wfSuppressWarnings();
 483                      $val = iconv( 'Windows-1252', 'UTF-8//IGNORE', $val );
 484                      wfRestoreWarnings();
 485                  }
 486              }
 487  
 488              //trim and check to make sure not only whitespace.
 489              $val = trim( $val );
 490              if ( strlen( $val ) === 0 ) {
 491                  //only whitespace.
 492                  $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, "$prop: Is only whitespace" );
 493                  unset( $this->mFilteredExifData[$prop] );
 494  
 495                  return;
 496              }
 497  
 498              //all's good.
 499              $this->mFilteredExifData[$prop] = $val;
 500          }
 501      }
 502  
 503      /**
 504       * Convert an Exif::UNDEFINED from a raw binary string
 505       * to its value. This is sometimes needed depending on
 506       * the type of UNDEFINED field
 507       * @param string $prop Name of property
 508       */
 509  	private function exifPropToOrd( $prop ) {
 510          if ( isset( $this->mFilteredExifData[$prop] ) ) {
 511              $this->mFilteredExifData[$prop] = ord( $this->mFilteredExifData[$prop] );
 512          }
 513      }
 514  
 515      /**
 516       * Convert gps in exif form to a single floating point number
 517       * for example 10 degress 20`40`` S -> -10.34444
 518       * @param string $prop A GPS coordinate exif tag name (like GPSLongitude)
 519       */
 520  	private function exifGPStoNumber( $prop ) {
 521          $loc =& $this->mFilteredExifData[$prop];
 522          $dir =& $this->mFilteredExifData[$prop . 'Ref'];
 523          $res = false;
 524  
 525          if ( isset( $loc ) && isset( $dir )
 526              && ( $dir === 'N' || $dir === 'S' || $dir === 'E' || $dir === 'W' )
 527          ) {
 528              list( $num, $denom ) = explode( '/', $loc[0] );
 529              $res = $num / $denom;
 530              list( $num, $denom ) = explode( '/', $loc[1] );
 531              $res += ( $num / $denom ) * ( 1 / 60 );
 532              list( $num, $denom ) = explode( '/', $loc[2] );
 533              $res += ( $num / $denom ) * ( 1 / 3600 );
 534  
 535              if ( $dir === 'S' || $dir === 'W' ) {
 536                  $res *= -1; // make negative
 537              }
 538          }
 539  
 540          // update the exif records.
 541  
 542          if ( $res !== false ) { // using !== as $res could potentially be 0
 543              $this->mFilteredExifData[$prop] = $res;
 544              unset( $this->mFilteredExifData[$prop . 'Ref'] );
 545          } else { // if invalid
 546              unset( $this->mFilteredExifData[$prop] );
 547              unset( $this->mFilteredExifData[$prop . 'Ref'] );
 548          }
 549      }
 550  
 551      /**#@-*/
 552  
 553      /**#@+
 554       * @return array
 555       */
 556      /**
 557       * Get $this->mRawExifData
 558       * @return array
 559       */
 560  	function getData() {
 561          return $this->mRawExifData;
 562      }
 563  
 564      /**
 565       * Get $this->mFilteredExifData
 566       * @return array
 567       */
 568  	function getFilteredData() {
 569          return $this->mFilteredExifData;
 570      }
 571  
 572      /**#@-*/
 573  
 574      /**
 575       * The version of the output format
 576       *
 577       * Before the actual metadata information is saved in the database we
 578       * strip some of it since we don't want to save things like thumbnails
 579       * which usually accompany Exif data. This value gets saved in the
 580       * database along with the actual Exif data, and if the version in the
 581       * database doesn't equal the value returned by this function the Exif
 582       * data is regenerated.
 583       *
 584       * @return int
 585       */
 586  	public static function version() {
 587          return 2; // We don't need no bloddy constants!
 588      }
 589  
 590      /**
 591       * Validates if a tag value is of the type it should be according to the Exif spec
 592       *
 593       * @param mixed $in The input value to check
 594       * @return bool
 595       */
 596  	private function isByte( $in ) {
 597          if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 255 ) {
 598              $this->debug( $in, __FUNCTION__, true );
 599  
 600              return true;
 601          } else {
 602              $this->debug( $in, __FUNCTION__, false );
 603  
 604              return false;
 605          }
 606      }
 607  
 608      /**
 609       * @param mixed $in The input value to check
 610       * @return bool
 611       */
 612  	private function isASCII( $in ) {
 613          if ( is_array( $in ) ) {
 614              return false;
 615          }
 616  
 617          if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 618              $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 619  
 620              return false;
 621          }
 622  
 623          if ( preg_match( '/^\s*$/', $in ) ) {
 624              $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 625  
 626              return false;
 627          }
 628  
 629          return true;
 630      }
 631  
 632      /**
 633       * @param mixed $in The input value to check
 634       * @return bool
 635       */
 636  	private function isShort( $in ) {
 637          if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 65536 ) {
 638              $this->debug( $in, __FUNCTION__, true );
 639  
 640              return true;
 641          } else {
 642              $this->debug( $in, __FUNCTION__, false );
 643  
 644              return false;
 645          }
 646      }
 647  
 648      /**
 649       * @param mixed $in The input value to check
 650       * @return bool
 651       */
 652  	private function isLong( $in ) {
 653          if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 4294967296 ) {
 654              $this->debug( $in, __FUNCTION__, true );
 655  
 656              return true;
 657          } else {
 658              $this->debug( $in, __FUNCTION__, false );
 659  
 660              return false;
 661          }
 662      }
 663  
 664      /**
 665       * @param mixed $in The input value to check
 666       * @return bool
 667       */
 668  	private function isRational( $in ) {
 669          $m = array();
 670  
 671          # Avoid division by zero
 672          if ( !is_array( $in )
 673              && preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 674          ) {
 675              return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 676          } else {
 677              $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 678  
 679              return false;
 680          }
 681      }
 682  
 683      /**
 684       * @param mixed $in The input value to check
 685       * @return bool
 686       */
 687  	private function isUndefined( $in ) {
 688          $this->debug( $in, __FUNCTION__, true );
 689  
 690          return true;
 691      }
 692  
 693      /**
 694       * @param mixed $in The input value to check
 695       * @return bool
 696       */
 697  	private function isSlong( $in ) {
 698          if ( $this->isLong( abs( $in ) ) ) {
 699              $this->debug( $in, __FUNCTION__, true );
 700  
 701              return true;
 702          } else {
 703              $this->debug( $in, __FUNCTION__, false );
 704  
 705              return false;
 706          }
 707      }
 708  
 709      /**
 710       * @param mixed $in The input value to check
 711       * @return bool
 712       */
 713  	private function isSrational( $in ) {
 714          $m = array();
 715  
 716          # Avoid division by zero
 717          if ( !is_array( $in ) &&
 718              preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 719          ) {
 720              return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 721          } else {
 722              $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 723  
 724              return false;
 725          }
 726      }
 727  
 728      /**#@-*/
 729  
 730      /**
 731       * Validates if a tag has a legal value according to the Exif spec
 732       *
 733       * @param string $section Section where tag is located.
 734       * @param string $tag The tag to check.
 735       * @param mixed $val The value of the tag.
 736       * @param bool $recursive True if called recursively for array types.
 737       * @return bool
 738       */
 739  	private function validate( $section, $tag, $val, $recursive = false ) {
 740          $debug = "tag is '$tag'";
 741          $etype = $this->mExifTags[$section][$tag];
 742          $ecount = 1;
 743          if ( is_array( $etype ) ) {
 744              list( $etype, $ecount ) = $etype;
 745              if ( $recursive ) {
 746                  $ecount = 1; // checking individual elements
 747              }
 748          }
 749          $count = count( $val );
 750          if ( $ecount != $count ) {
 751              $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 752  
 753              return false;
 754          }
 755          if ( $count > 1 ) {
 756              foreach ( $val as $v ) {
 757                  if ( !$this->validate( $section, $tag, $v, true ) ) {
 758                      return false;
 759                  }
 760              }
 761  
 762              return true;
 763          }
 764          // Does not work if not typecast
 765          switch ( (string)$etype ) {
 766              case (string)Exif::BYTE:
 767                  $this->debug( $val, __FUNCTION__, $debug );
 768  
 769                  return $this->isByte( $val );
 770              case (string)Exif::ASCII:
 771                  $this->debug( $val, __FUNCTION__, $debug );
 772  
 773                  return $this->isASCII( $val );
 774              case (string)Exif::SHORT:
 775                  $this->debug( $val, __FUNCTION__, $debug );
 776  
 777                  return $this->isShort( $val );
 778              case (string)Exif::LONG:
 779                  $this->debug( $val, __FUNCTION__, $debug );
 780  
 781                  return $this->isLong( $val );
 782              case (string)Exif::RATIONAL:
 783                  $this->debug( $val, __FUNCTION__, $debug );
 784  
 785                  return $this->isRational( $val );
 786              case (string)Exif::SHORT_OR_LONG:
 787                  $this->debug( $val, __FUNCTION__, $debug );
 788  
 789                  return $this->isShort( $val ) || $this->isLong( $val );
 790              case (string)Exif::UNDEFINED:
 791                  $this->debug( $val, __FUNCTION__, $debug );
 792  
 793                  return $this->isUndefined( $val );
 794              case (string)Exif::SLONG:
 795                  $this->debug( $val, __FUNCTION__, $debug );
 796  
 797                  return $this->isSlong( $val );
 798              case (string)Exif::SRATIONAL:
 799                  $this->debug( $val, __FUNCTION__, $debug );
 800  
 801                  return $this->isSrational( $val );
 802              case (string)Exif::IGNORE:
 803                  $this->debug( $val, __FUNCTION__, $debug );
 804  
 805                  return false;
 806              default:
 807                  $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 808  
 809                  return false;
 810          }
 811      }
 812  
 813      /**
 814       * Convenience function for debugging output
 815       *
 816       * @param mixed $in Arrays will be processed with print_r().
 817       * @param string $fname Function name to log.
 818       * @param string|bool|null $action Default null.
 819       */
 820  	private function debug( $in, $fname, $action = null ) {
 821          if ( !$this->log ) {
 822              return;
 823          }
 824          $type = gettype( $in );
 825          $class = ucfirst( __CLASS__ );
 826          if ( is_array( $in ) ) {
 827              $in = print_r( $in, true );
 828          }
 829  
 830          if ( $action === true ) {
 831              wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)" );
 832          } elseif ( $action === false ) {
 833              wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)" );
 834          } elseif ( $action === null ) {
 835              wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)" );
 836          } else {
 837              wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')" );
 838          }
 839      }
 840  
 841      /**
 842       * Convenience function for debugging output
 843       *
 844       * @param string $fname The name of the function calling this function
 845       * @param bool $io Specify whether we're beginning or ending
 846       */
 847  	private function debugFile( $fname, $io ) {
 848          if ( !$this->log ) {
 849              return;
 850          }
 851          $class = ucfirst( __CLASS__ );
 852          if ( $io ) {
 853              wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'" );
 854          } else {
 855              wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'" );
 856          }
 857      }
 858  }