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         const BYTE = 1; //!< An 8-bit (1-byte) unsigned integer.
  34         const ASCII = 2; //!< An 8-bit byte containing one 7-bit ASCII code. The final byte is terminated with NULL.
  35         const SHORT = 3; //!< A 16-bit (2-byte) unsigned integer.
  36         const LONG = 4; //!< A 32-bit (4-byte) unsigned integer.
  37         const RATIONAL = 5; //!< Two LONGs. The first LONG is the numerator and the second LONG expresses the denominator
  38         const SHORT_OR_LONG = 6; //!< A 16-bit (2-byte) or 32-bit (4-byte) unsigned integer.
  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         /**
 104          * Constructor
 105          *
 106          * @param string $file filename.
 107          * @param string $byteOrder Type of byte ordering either 'BE' (Big Endian) or 'LE' (Little Endian). Default ''.
 108          * @throws MWException
 109          * @todo FIXME: The following are broke:
 110          * SubjectArea. Need to test the more obscure tags.
 111          *
 112          * DigitalZoomRatio = 0/0 is rejected. need to determine if that's valid.
 113          * possibly should treat 0/0 = 0. need to read exif spec on that.
 114          */
 115         function __construct( $file, $byteOrder = '' ) {
 116                 /**
 117                  * Page numbers here refer to pages in the Exif 2.2 standard
 118                  *
 119                  * Note, Exif::UNDEFINED is treated as a string, not as an array of bytes
 120                  * so don't put a count parameter for any UNDEFINED values.
 121                  *
 122                  * @link http://exif.org/Exif2-2.PDF The Exif 2.2 specification
 123                  */
 124                 $this->mExifTags = array(
 125                         # TIFF Rev. 6.0 Attribute Information (p22)
 126                         'IFD0' => array(
 127                                 # Tags relating to image structure
 128                                 'ImageWidth' => Exif::SHORT_OR_LONG, # Image width
 129                                 'ImageLength' => Exif::SHORT_OR_LONG, # Image height
 130                                 'BitsPerSample' => array( Exif::SHORT, 3 ), # Number of bits per component
 131                                 # "When a primary image is JPEG compressed, this designation is not"
 132                                 # "necessary and is omitted." (p23)
 133                                 'Compression' => Exif::SHORT, # Compression scheme #p23
 134                                 'PhotometricInterpretation' => Exif::SHORT, # Pixel composition #p23
 135                                 'Orientation' => Exif::SHORT, # Orientation of image #p24
 136                                 'SamplesPerPixel' => Exif::SHORT, # Number of components
 137                                 'PlanarConfiguration' => Exif::SHORT, # Image data arrangement #p24
 138                                 'YCbCrSubSampling' => array( Exif::SHORT, 2 ), # Subsampling ratio of Y to C #p24
 139                                 'YCbCrPositioning' => Exif::SHORT, # Y and C positioning #p24-25
 140                                 'XResolution' => Exif::RATIONAL, # Image resolution in width direction
 141                                 'YResolution' => Exif::RATIONAL, # Image resolution in height direction
 142                                 'ResolutionUnit' => Exif::SHORT, # Unit of X and Y resolution #(p26)
 143
 144                                 # Tags relating to recording offset
 145                                 'StripOffsets' => Exif::SHORT_OR_LONG, # Image data location
 146                                 'RowsPerStrip' => Exif::SHORT_OR_LONG, # Number of rows per strip
 147                                 'StripByteCounts' => Exif::SHORT_OR_LONG, # Bytes per compressed strip
 148                                 'JPEGInterchangeFormat' => Exif::SHORT_OR_LONG, # Offset to JPEG SOI
 149                                 'JPEGInterchangeFormatLength' => Exif::SHORT_OR_LONG, # Bytes of JPEG data
 150
 151                                 # Tags relating to image data characteristics
 152                                 'TransferFunction' => Exif::IGNORE, # Transfer function
 153                                 'WhitePoint' => array( Exif::RATIONAL, 2 ), # White point chromaticity
 154                                 'PrimaryChromaticities' => array( Exif::RATIONAL, 6 ), # Chromaticities of primarities
 155                                 'YCbCrCoefficients' => array( Exif::RATIONAL, 3 ), # Color space transformation matrix coefficients #p27
 156                                 'ReferenceBlackWhite' => array( Exif::RATIONAL, 6 ), # Pair of black and white reference values
 157
 158                                 # Other tags
 159                                 'DateTime' => Exif::ASCII, # File change date and time
 160                                 'ImageDescription' => Exif::ASCII, # Image title
 161                                 'Make' => Exif::ASCII, # Image input equipment manufacturer
 162                                 'Model' => Exif::ASCII, # Image input equipment model
 163                                 'Software' => Exif::ASCII, # Software used
 164                                 'Artist' => Exif::ASCII, # Person who created the image
 165                                 'Copyright' => Exif::ASCII, # Copyright holder
 166                         ),
 167
 168                         # Exif IFD Attribute Information (p30-31)
 169                         'EXIF' => array(
 170                                 # TODO: NOTE: Nonexistence of this field is taken to mean nonconformance
 171                                 # to the Exif 2.1 AND 2.2 standards
 172                                 'ExifVersion' => Exif::UNDEFINED, # Exif version
 173                                 'FlashPixVersion' => Exif::UNDEFINED, # Supported Flashpix version #p32
 174
 175                                 # Tags relating to Image Data Characteristics
 176                                 'ColorSpace' => Exif::SHORT, # Color space information #p32
 177
 178                                 # Tags relating to image configuration
 179                                 'ComponentsConfiguration' => Exif::UNDEFINED, # Meaning of each component #p33
 180                                 'CompressedBitsPerPixel' => Exif::RATIONAL, # Image compression mode
 181                                 'PixelYDimension' => Exif::SHORT_OR_LONG, # Valid image width
 182                                 'PixelXDimension' => Exif::SHORT_OR_LONG, # Valid image height
 183
 184                                 # Tags relating to related user information
 185                                 'MakerNote' => Exif::IGNORE, # Manufacturer notes
 186                                 'UserComment' => Exif::UNDEFINED, # User comments #p34
 187
 188                                 # Tags relating to related file information
 189                                 'RelatedSoundFile' => Exif::ASCII, # Related audio file
 190
 191                                 # Tags relating to date and time
 192                                 'DateTimeOriginal' => Exif::ASCII, # Date and time of original data generation #p36
 193                                 'DateTimeDigitized' => Exif::ASCII, # Date and time of original data generation
 194                                 'SubSecTime' => Exif::ASCII, # DateTime subseconds
 195                                 'SubSecTimeOriginal' => Exif::ASCII, # DateTimeOriginal subseconds
 196                                 'SubSecTimeDigitized' => Exif::ASCII, # DateTimeDigitized subseconds
 197
 198                                 # Tags relating to picture-taking conditions (p31)
 199                                 'ExposureTime' => Exif::RATIONAL, # Exposure time
 200                                 'FNumber' => Exif::RATIONAL, # F Number
 201                                 'ExposureProgram' => Exif::SHORT, # Exposure Program #p38
 202                                 'SpectralSensitivity' => Exif::ASCII, # Spectral sensitivity
 203                                 'ISOSpeedRatings' => Exif::SHORT, # ISO speed rating
 204                                 'OECF' => Exif::IGNORE,
 205                                 # Optoelectronic conversion factor. Note: We don't have support for this atm.
 206                                 'ShutterSpeedValue' => Exif::SRATIONAL, # Shutter speed
 207                                 'ApertureValue' => Exif::RATIONAL, # Aperture
 208                                 'BrightnessValue' => Exif::SRATIONAL, # Brightness
 209                                 'ExposureBiasValue' => Exif::SRATIONAL, # Exposure bias
 210                                 'MaxApertureValue' => Exif::RATIONAL, # Maximum land aperture
 211                                 'SubjectDistance' => Exif::RATIONAL, # Subject distance
 212                                 'MeteringMode' => Exif::SHORT, # Metering mode #p40
 213                                 'LightSource' => Exif::SHORT, # Light source #p40-41
 214                                 'Flash' => Exif::SHORT, # Flash #p41-42
 215                                 'FocalLength' => Exif::RATIONAL, # Lens focal length
 216                                 'SubjectArea' => array( Exif::SHORT, 4 ), # Subject area
 217                                 'FlashEnergy' => Exif::RATIONAL, # Flash energy
 218                                 'SpatialFrequencyResponse' => Exif::IGNORE, # Spatial frequency response. Not supported atm.
 219                                 'FocalPlaneXResolution' => Exif::RATIONAL, # Focal plane X resolution
 220                                 'FocalPlaneYResolution' => Exif::RATIONAL, # Focal plane Y resolution
 221                                 'FocalPlaneResolutionUnit' => Exif::SHORT, # Focal plane resolution unit #p46
 222                                 'SubjectLocation' => array( Exif::SHORT, 2 ), # Subject location
 223                                 'ExposureIndex' => Exif::RATIONAL, # Exposure index
 224                                 'SensingMethod' => Exif::SHORT, # Sensing method #p46
 225                                 'FileSource' => Exif::UNDEFINED, # File source #p47
 226                                 'SceneType' => Exif::UNDEFINED, # Scene type #p47
 227                                 'CFAPattern' => Exif::IGNORE, # CFA pattern. not supported atm.
 228                                 'CustomRendered' => Exif::SHORT, # Custom image processing #p48
 229                                 'ExposureMode' => Exif::SHORT, # Exposure mode #p48
 230                                 'WhiteBalance' => Exif::SHORT, # White Balance #p49
 231                                 'DigitalZoomRatio' => Exif::RATIONAL, # Digital zoom ration
 232                                 'FocalLengthIn35mmFilm' => Exif::SHORT, # Focal length in 35 mm film
 233                                 'SceneCaptureType' => Exif::SHORT, # Scene capture type #p49
 234                                 'GainControl' => Exif::SHORT, # Scene control #p49-50
 235                                 'Contrast' => Exif::SHORT, # Contrast #p50
 236                                 'Saturation' => Exif::SHORT, # Saturation #p50
 237                                 'Sharpness' => Exif::SHORT, # Sharpness #p50
 238                                 'DeviceSettingDescription' => Exif::IGNORE,
 239                                 # Device settings description. This could maybe be supported. Need to find an
 240                                 # example file that uses this to see if it has stuff of interest in it.
 241                                 'SubjectDistanceRange' => Exif::SHORT, # Subject distance range #p51
 242
 243                                 'ImageUniqueID' => Exif::ASCII, # Unique image ID
 244                         ),
 245
 246                         # GPS Attribute Information (p52)
 247                         'GPS' => array(
 248                                 'GPSVersion' => Exif::UNDEFINED,
 249                                 # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 250                                 # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 251                                 'GPSLatitudeRef' => Exif::ASCII, # North or South Latitude #p52-53
 252                                 'GPSLatitude' => array( Exif::RATIONAL, 3 ), # Latitude
 253                                 'GPSLongitudeRef' => Exif::ASCII, # East or West Longitude #p53
 254                                 'GPSLongitude' => array( Exif::RATIONAL, 3 ), # Longitude
 255                                 'GPSAltitudeRef' => Exif::UNDEFINED,
 256                                 # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 257                                 # but php seems to disagree.
 258                                 'GPSAltitude' => Exif::RATIONAL, # Altitude
 259                                 'GPSTimeStamp' => array( Exif::RATIONAL, 3 ), # GPS time (atomic clock)
 260                                 'GPSSatellites' => Exif::ASCII, # Satellites used for measurement
 261                                 'GPSStatus' => Exif::ASCII, # Receiver status #p54
 262                                 'GPSMeasureMode' => Exif::ASCII, # Measurement mode #p54-55
 263                                 'GPSDOP' => Exif::RATIONAL, # Measurement precision
 264                                 'GPSSpeedRef' => Exif::ASCII, # Speed unit #p55
 265                                 'GPSSpeed' => Exif::RATIONAL, # Speed of GPS receiver
 266                                 'GPSTrackRef' => Exif::ASCII, # Reference for direction of movement #p55
 267                                 'GPSTrack' => Exif::RATIONAL, # Direction of movement
 268                                 'GPSImgDirectionRef' => Exif::ASCII, # Reference for direction of image #p56
 269                                 'GPSImgDirection' => Exif::RATIONAL, # Direction of image
 270                                 'GPSMapDatum' => Exif::ASCII, # Geodetic survey data used
 271                                 'GPSDestLatitudeRef' => Exif::ASCII, # Reference for latitude of destination #p56
 272                                 'GPSDestLatitude' => array( Exif::RATIONAL, 3 ), # Latitude destination
 273                                 'GPSDestLongitudeRef' => Exif::ASCII, # Reference for longitude of destination #p57
 274                                 'GPSDestLongitude' => array( Exif::RATIONAL, 3 ), # Longitude of destination
 275                                 'GPSDestBearingRef' => Exif::ASCII, # Reference for bearing of destination #p57
 276                                 'GPSDestBearing' => Exif::RATIONAL, # Bearing of destination
 277                                 'GPSDestDistanceRef' => Exif::ASCII, # Reference for distance to destination #p57-58
 278                                 'GPSDestDistance' => Exif::RATIONAL, # Distance to destination
 279                                 'GPSProcessingMethod' => Exif::UNDEFINED, # Name of GPS processing method
 280                                 'GPSAreaInformation' => Exif::UNDEFINED, # Name of GPS area
 281                                 'GPSDateStamp' => Exif::ASCII, # GPS date
 282                                 'GPSDifferential' => Exif::SHORT, # GPS differential correction
 283                         ),
 284                 );
 285
 286                 $this->file = $file;
 287                 $this->basename = wfBaseName( $this->file );
 288                 if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 289                         $this->byteOrder = $byteOrder;
 290                 } else {
 291                         // Only give a warning for b/c, since originally we didn't
 292                         // require this. The number of things affected by this is
 293                         // rather small.
 294                         wfWarn( 'Exif class did not have byte order specified. ' .
 295                                 'Some properties may be decoded incorrectly.' );
 296                         $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 297                 }
 298
 299                 $this->debugFile( $this->basename, __FUNCTION__, true );
 300                 if ( function_exists( 'exif_read_data' ) ) {
 301                         wfSuppressWarnings();
 302                         $data = exif_read_data( $this->file, 0, true );
 303                         wfRestoreWarnings();
 304                 } else {
 305                         throw new MWException( "Internal error: exif_read_data not present. \$wgShowEXIF may be incorrectly set or not checked by an extension." );
 306                 }
 307                 /**
 308                  * exif_read_data() will return false on invalid input, such as
 309                  * when somebody uploads a file called something.jpeg
 310                  * containing random gibberish.
 311                  */
 312                 $this->mRawExifData = $data ? $data : array();
 313                 $this->makeFilteredData();
 314                 $this->collapseData();
 315                 $this->debugFile( __FUNCTION__, false );
 316         }
 317
 318         /**
 319          * Make $this->mFilteredExifData
 320          */
 321         function makeFilteredData() {
 322                 $this->mFilteredExifData = array();
 323
 324                 foreach ( array_keys( $this->mRawExifData ) as $section ) {
 325                         if ( !in_array( $section, array_keys( $this->mExifTags ) ) ) {
 326                                 $this->debug( $section, __FUNCTION__, "'$section' is not a valid Exif section" );
 327                                 continue;
 328                         }
 329
 330                         foreach ( array_keys( $this->mRawExifData[$section] ) as $tag ) {
 331                                 if ( !in_array( $tag, array_keys( $this->mExifTags[$section] ) ) ) {
 332                                         $this->debug( $tag, __FUNCTION__, "'$tag' is not a valid tag in '$section'" );
 333                                         continue;
 334                                 }
 335
 336                                 $this->mFilteredExifData[$tag] = $this->mRawExifData[$section][$tag];
 337                                 // This is ok, as the tags in the different sections do not conflict.
 338                                 // except in computed and thumbnail section, which we don't use.
 339
 340                                 $value = $this->mRawExifData[$section][$tag];
 341                                 if ( !$this->validate( $section, $tag, $value ) ) {
 342                                         $this->debug( $value, __FUNCTION__, "'$tag' contained invalid data" );
 343                                         unset( $this->mFilteredExifData[$tag] );
 344                                 }
 345                         }
 346                 }
 347         }
 348
 349         /**
 350          * Collapse some fields together.
 351          * This converts some fields from exif form, to a more friendly form.
 352          * For example GPS latitude to a single number.
 353          *
 354          * The rationale behind this is that we're storing data, not presenting to the user
 355          * For example a longitude is a single number describing how far away you are from
 356          * the prime meridian. Well it might be nice to split it up into minutes and seconds
 357          * for the user, it doesn't really make sense to split a single number into 4 parts
 358          * for storage. (degrees, minutes, second, direction vs single floating point number).
 359          *
 360          * Other things this might do (not really sure if they make sense or not):
 361          * Dates -> mediawiki date format.
 362          * convert values that can be in different units to be in one standardized unit.
 363          *
 364          * As an alternative approach, some of this could be done in the validate phase
 365          * if we make up our own types like Exif::DATE.
 366          */
 367         function collapseData() {
 368
 369                 $this->exifGPStoNumber( 'GPSLatitude' );
 370                 $this->exifGPStoNumber( 'GPSDestLatitude' );
 371                 $this->exifGPStoNumber( 'GPSLongitude' );
 372                 $this->exifGPStoNumber( 'GPSDestLongitude' );
 373
 374                 if ( isset( $this->mFilteredExifData['GPSAltitude'] ) && isset( $this->mFilteredExifData['GPSAltitudeRef'] ) ) {
 375
 376                         // We know altitude data is a <num>/<denom> from the validation functions ran earlier.
 377                         // But multiplying such a string by -1 doesn't work well, so convert.
 378                         list( $num, $denom ) = explode( '/', $this->mFilteredExifData['GPSAltitude'] );
 379                         $this->mFilteredExifData['GPSAltitude'] = $num / $denom;
 380
 381                         if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 382                                 $this->mFilteredExifData['GPSAltitude'] *= -1;
 383                         }
 384                         unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 385                 }
 386
 387                 $this->exifPropToOrd( 'FileSource' );
 388                 $this->exifPropToOrd( 'SceneType' );
 389
 390                 $this->charCodeString( 'UserComment' );
 391                 $this->charCodeString( 'GPSProcessingMethod' );
 392                 $this->charCodeString( 'GPSAreaInformation' );
 393
 394                 //ComponentsConfiguration should really be an array instead of a string...
 395                 //This turns a string of binary numbers into an array of numbers.
 396
 397                 if ( isset( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 398                         $val = $this->mFilteredExifData['ComponentsConfiguration'];
 399                         $ccVals = array();
 400                         for ( $i = 0; $i < strlen( $val ); $i++ ) {
 401                                 $ccVals[$i] = ord( substr( $val, $i, 1 ) );
 402                         }
 403                         $ccVals['_type'] = 'ol'; //this is for formatting later.
 404                         $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 405                 }
 406
 407                 //GPSVersion(ID) is treated as the wrong type by php exif support.
 408                 //Go through each byte turning it into a version string.
 409                 //For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 410
 411                 //Also change exif tag name from GPSVersion (what php exif thinks it is)
 412                 //to GPSVersionID (what the exif standard thinks it is).
 413
 414                 if ( isset( $this->mFilteredExifData['GPSVersion'] ) ) {
 415                         $val = $this->mFilteredExifData['GPSVersion'];
 416                         $newVal = '';
 417                         for ( $i = 0; $i < strlen( $val ); $i++ ) {
 418                                 if ( $i !== 0 ) {
 419                                         $newVal .= '.';
 420                                 }
 421                                 $newVal .= ord( substr( $val, $i, 1 ) );
 422                         }
 423                         if ( $this->byteOrder === 'LE' ) {
 424                                 // Need to reverse the string
 425                                 $newVal2 = '';
 426                                 for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 427                                         $newVal2 .= substr( $newVal, $i, 1 );
 428                                 }
 429                                 $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 430                         } else {
 431                                 $this->mFilteredExifData['GPSVersionID'] = $newVal;
 432                         }
 433                         unset( $this->mFilteredExifData['GPSVersion'] );
 434                 }
 435         }
 436
 437         /**
 438          * Do userComment tags and similar. See pg. 34 of exif standard.
 439          * basically first 8 bytes is charset, rest is value.
 440          * This has not been tested on any shift-JIS strings.
 441          * @param string $prop prop name.
 442          */
 443         private function charCodeString( $prop ) {
 444                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 445
 446                         if ( strlen( $this->mFilteredExifData[$prop] ) <= 8 ) {
 447                                 //invalid. Must be at least 9 bytes long.
 448
 449                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, false );
 450                                 unset( $this->mFilteredExifData[$prop] );
 451
 452                                 return;
 453                         }
 454                         $charCode = substr( $this->mFilteredExifData[$prop], 0, 8 );
 455                         $val = substr( $this->mFilteredExifData[$prop], 8 );
 456
 457                         switch ( $charCode ) {
 458                                 case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 459                                         //JIS
 460                                         $charset = "Shift-JIS";
 461                                         break;
 462                                 case "UNICODE\x00":
 463                                         $charset = "UTF-16" . $this->byteOrder;
 464                                         break;
 465                                 default: //ascii or undefined.
 466                                         $charset = "";
 467                                         break;
 468                         }
 469                         // This could possibly check to see if iconv is really installed
 470                         // or if we're using the compatibility wrapper in globalFunctions.php
 471                         if ( $charset ) {
 472                                 wfSuppressWarnings();
 473                                 $val = iconv( $charset, 'UTF-8//IGNORE', $val );
 474                                 wfRestoreWarnings();
 475                         } else {
 476                                 // if valid utf-8, assume that, otherwise assume windows-1252
 477                                 $valCopy = $val;
 478                                 UtfNormal::quickIsNFCVerify( $valCopy ); //validates $valCopy.
 479                                 if ( $valCopy !== $val ) {
 480                                         wfSuppressWarnings();
 481                                         $val = iconv( 'Windows-1252', 'UTF-8//IGNORE', $val );
 482                                         wfRestoreWarnings();
 483                                 }
 484                         }
 485
 486                         //trim and check to make sure not only whitespace.
 487                         $val = trim( $val );
 488                         if ( strlen( $val ) === 0 ) {
 489                                 //only whitespace.
 490                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, "$prop: Is only whitespace" );
 491                                 unset( $this->mFilteredExifData[$prop] );
 492
 493                                 return;
 494                         }
 495
 496                         //all's good.
 497                         $this->mFilteredExifData[$prop] = $val;
 498                 }
 499         }
 500
 501         /**
 502          * Convert an Exif::UNDEFINED from a raw binary string
 503          * to its value. This is sometimes needed depending on
 504          * the type of UNDEFINED field
 505          * @param string $prop name of property
 506          */
 507         private function exifPropToOrd( $prop ) {
 508                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 509                         $this->mFilteredExifData[$prop] = ord( $this->mFilteredExifData[$prop] );
 510                 }
 511         }
 512
 513         /**
 514          * Convert gps in exif form to a single floating point number
 515          * for example 10 degress 20`40`` S -> -10.34444
 516          * @param string $prop a gps coordinate exif tag name (like GPSLongitude)
 517          */
 518         private function exifGPStoNumber( $prop ) {
 519                 $loc =& $this->mFilteredExifData[$prop];
 520                 $dir =& $this->mFilteredExifData[$prop . 'Ref'];
 521                 $res = false;
 522
 523                 if ( isset( $loc ) && isset( $dir ) && ( $dir === 'N' || $dir === 'S' || $dir === 'E' || $dir === 'W' ) ) {
 524                         list( $num, $denom ) = explode( '/', $loc[0] );
 525                         $res = $num / $denom;
 526                         list( $num, $denom ) = explode( '/', $loc[1] );
 527                         $res += ( $num / $denom ) * ( 1 / 60 );
 528                         list( $num, $denom ) = explode( '/', $loc[2] );
 529                         $res += ( $num / $denom ) * ( 1 / 3600 );
 530
 531                         if ( $dir === 'S' || $dir === 'W' ) {
 532                                 $res *= -1; // make negative
 533                         }
 534                 }
 535
 536                 // update the exif records.
 537
 538                 if ( $res !== false ) { // using !== as $res could potentially be 0
 539                         $this->mFilteredExifData[$prop] = $res;
 540                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 541                 } else { // if invalid
 542                         unset( $this->mFilteredExifData[$prop] );
 543                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 544                 }
 545         }
 546
 547         /**
 548          * Use FormatMetadata to create formatted values for display to user
 549          * (is this ever used?)
 550          *
 551          * @deprecated since 1.18
 552          */
 553         function makeFormattedData() {
 554                 wfDeprecated( __METHOD__, '1.18' );
 555                 $this->mFormattedExifData = FormatMetadata::getFormattedData(
 556                         $this->mFilteredExifData );
 557         }
 558
 559         /**#@-*/
 560
 561         /**#@+
 562          * @return array
 563          */
 564         /**
 565          * Get $this->mRawExifData
 566          * @return array
 567          */
 568         function getData() {
 569                 return $this->mRawExifData;
 570         }
 571
 572         /**
 573          * Get $this->mFilteredExifData
 574          */
 575         function getFilteredData() {
 576                 return $this->mFilteredExifData;
 577         }
 578
 579         /**
 580          * Get $this->mFormattedExifData
 581          *
 582          * This returns the data for display to user.
 583          * Its unclear if this is ever used.
 584          *
 585          * @deprecated since 1.18
 586          */
 587         function getFormattedData() {
 588                 wfDeprecated( __METHOD__, '1.18' );
 589                 if ( !$this->mFormattedExifData ) {
 590                         $this->makeFormattedData();
 591                 }
 592
 593                 return $this->mFormattedExifData;
 594         }
 595
 596         /**#@-*/
 597
 598         /**
 599          * The version of the output format
 600          *
 601          * Before the actual metadata information is saved in the database we
 602          * strip some of it since we don't want to save things like thumbnails
 603          * which usually accompany Exif data. This value gets saved in the
 604          * database along with the actual Exif data, and if the version in the
 605          * database doesn't equal the value returned by this function the Exif
 606          * data is regenerated.
 607          *
 608          * @return int
 609          */
 610         public static function version() {
 611                 return 2; // We don't need no bloddy constants!
 612         }
 613
 614         /**#@+
 615          * Validates if a tag value is of the type it should be according to the Exif spec
 616          *
 617          * @private
 618          *
 619          * @param $in Mixed: the input value to check
 620          * @return bool
 621          */
 622         private function isByte( $in ) {
 623                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 255 ) {
 624                         $this->debug( $in, __FUNCTION__, true );
 625
 626                         return true;
 627                 } else {
 628                         $this->debug( $in, __FUNCTION__, false );
 629
 630                         return false;
 631                 }
 632         }
 633
 634         /**
 635          * @param $in
 636          * @return bool
 637          */
 638         private function isASCII( $in ) {
 639                 if ( is_array( $in ) ) {
 640                         return false;
 641                 }
 642
 643                 if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 644                         $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 645
 646                         return false;
 647                 }
 648
 649                 if ( preg_match( '/^\s*$/', $in ) ) {
 650                         $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 651
 652                         return false;
 653                 }
 654
 655                 return true;
 656         }
 657
 658         /**
 659          * @param $in
 660          * @return bool
 661          */
 662         private function isShort( $in ) {
 663                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 65536 ) {
 664                         $this->debug( $in, __FUNCTION__, true );
 665
 666                         return true;
 667                 } else {
 668                         $this->debug( $in, __FUNCTION__, false );
 669
 670                         return false;
 671                 }
 672         }
 673
 674         /**
 675          * @param $in
 676          * @return bool
 677          */
 678         private function isLong( $in ) {
 679                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 4294967296 ) {
 680                         $this->debug( $in, __FUNCTION__, true );
 681
 682                         return true;
 683                 } else {
 684                         $this->debug( $in, __FUNCTION__, false );
 685
 686                         return false;
 687                 }
 688         }
 689
 690         /**
 691          * @param $in
 692          * @return bool
 693          */
 694         private function isRational( $in ) {
 695                 $m = array();
 696                 if ( !is_array( $in ) && preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m ) ) { # Avoid division by zero
 697                         return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 698                 } else {
 699                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 700
 701                         return false;
 702                 }
 703         }
 704
 705         /**
 706          * @param $in
 707          * @return bool
 708          */
 709         private function isUndefined( $in ) {
 710                 $this->debug( $in, __FUNCTION__, true );
 711
 712                 return true;
 713         }
 714
 715         /**
 716          * @param $in
 717          * @return bool
 718          */
 719         private function isSlong( $in ) {
 720                 if ( $this->isLong( abs( $in ) ) ) {
 721                         $this->debug( $in, __FUNCTION__, true );
 722
 723                         return true;
 724                 } else {
 725                         $this->debug( $in, __FUNCTION__, false );
 726
 727                         return false;
 728                 }
 729         }
 730
 731         /**
 732          * @param $in
 733          * @return bool
 734          */
 735         private function isSrational( $in ) {
 736                 $m = array();
 737                 if ( !is_array( $in ) && preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m ) ) { # Avoid division by zero
 738                         return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 739                 } else {
 740                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 741
 742                         return false;
 743                 }
 744         }
 745
 746         /**#@-*/
 747
 748         /**
 749          * Validates if a tag has a legal value according to the Exif spec
 750          *
 751          * @private
 752          * @param string $section section where tag is located.
 753          * @param string $tag the tag to check.
 754          * @param $val Mixed: the value of the tag.
 755          * @param $recursive Boolean: true if called recursively for array types.
 756          * @return bool
 757          */
 758         private function validate( $section, $tag, $val, $recursive = false ) {
 759                 $debug = "tag is '$tag'";
 760                 $etype = $this->mExifTags[$section][$tag];
 761                 $ecount = 1;
 762                 if ( is_array( $etype ) ) {
 763                         list( $etype, $ecount ) = $etype;
 764                         if ( $recursive ) {
 765                                 $ecount = 1; // checking individual elements
 766                         }
 767                 }
 768                 $count = count( $val );
 769                 if ( $ecount != $count ) {
 770                         $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 771
 772                         return false;
 773                 }
 774                 if ( $count > 1 ) {
 775                         foreach ( $val as $v ) {
 776                                 if ( !$this->validate( $section, $tag, $v, true ) ) {
 777                                         return false;
 778                                 }
 779                         }
 780
 781                         return true;
 782                 }
 783                 // Does not work if not typecast
 784                 switch ( (string)$etype ) {
 785                         case (string)Exif::BYTE:
 786                                 $this->debug( $val, __FUNCTION__, $debug );
 787
 788                                 return $this->isByte( $val );
 789                         case (string)Exif::ASCII:
 790                                 $this->debug( $val, __FUNCTION__, $debug );
 791
 792                                 return $this->isASCII( $val );
 793                         case (string)Exif::SHORT:
 794                                 $this->debug( $val, __FUNCTION__, $debug );
 795
 796                                 return $this->isShort( $val );
 797                         case (string)Exif::LONG:
 798                                 $this->debug( $val, __FUNCTION__, $debug );
 799
 800                                 return $this->isLong( $val );
 801                         case (string)Exif::RATIONAL:
 802                                 $this->debug( $val, __FUNCTION__, $debug );
 803
 804                                 return $this->isRational( $val );
 805                         case (string)Exif::SHORT_OR_LONG:
 806                                 $this->debug( $val, __FUNCTION__, $debug );
 807
 808                                 return $this->isShort( $val ) || $this->isLong( $val );
 809                         case (string)Exif::UNDEFINED:
 810                                 $this->debug( $val, __FUNCTION__, $debug );
 811
 812                                 return $this->isUndefined( $val );
 813                         case (string)Exif::SLONG:
 814                                 $this->debug( $val, __FUNCTION__, $debug );
 815
 816                                 return $this->isSlong( $val );
 817                         case (string)Exif::SRATIONAL:
 818                                 $this->debug( $val, __FUNCTION__, $debug );
 819
 820                                 return $this->isSrational( $val );
 821                         case (string)Exif::IGNORE:
 822                                 $this->debug( $val, __FUNCTION__, $debug );
 823
 824                                 return false;
 825                         default:
 826                                 $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 827
 828                                 return false;
 829                 }
 830         }
 831
 832         /**
 833          * Convenience function for debugging output
 834          *
 835          * @private
 836          *
 837          * @param $in Mixed:
 838          * @param $fname String:
 839          * @param $action Mixed: , default NULL.
 840          */
 841         private function debug( $in, $fname, $action = null ) {
 842                 if ( !$this->log ) {
 843                         return;
 844                 }
 845                 $type = gettype( $in );
 846                 $class = ucfirst( __CLASS__ );
 847                 if ( is_array( $in ) ) {
 848                         $in = print_r( $in, true );
 849                 }
 850
 851                 if ( $action === true ) {
 852                         wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)\n" );
 853                 } elseif ( $action === false ) {
 854                         wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)\n" );
 855                 } elseif ( $action === null ) {
 856                         wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)\n" );
 857                 } else {
 858                         wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')\n" );
 859                 }
 860         }
 861
 862         /**
 863          * Convenience function for debugging output
 864          *
 865          * @private
 866          *
 867          * @param string $fname the name of the function calling this function
 868          * @param $io Boolean: Specify whether we're beginning or ending
 869          */
 870         private function debugFile( $fname, $io ) {
 871                 if ( !$this->log ) {
 872                         return;
 873                 }
 874                 $class = ucfirst( __CLASS__ );
 875                 if ( $io ) {
 876                         wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'\n" );
 877                 } else {
 878                         wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'\n" );
 879                 }
 880         }
 881 }