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         /** 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 array Filtered and formatted Exif data, see FormatMetadata::getFormattedData() */
  85         private $mFormattedExifData;
  86
  87         /** @var string The file being processed */
  88         private $file;
  89
  90         /** @var string The basename of the file being processed */
  91         private $basename;
  92
  93         /** @var string The private log to log to, e.g. 'exif' */
  94         private $log = false;
  95
  96         /** @var string The byte order of the file. Needed because php's extension
  97          *    doesn't fully process some obscure props.
  98          */
  99         private $byteOrder;
 100
 101         /**
 102          * Constructor
 103          *
 104          * @param string $file Filename.
 105          * @param string $byteOrder Type of byte ordering either 'BE' (Big Endian)
 106          *   or 'LE' (Little Endian). Default ''.
 107          * @throws MWException
 108          * @todo FIXME: The following are broke:
 109          *   SubjectArea. Need to test the more obscure tags.
 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_OR_LONG, # Image width
 127                                 'ImageLength' => Exif::SHORT_OR_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_OR_LONG, # Image data location
 144                                 'RowsPerStrip' => Exif::SHORT_OR_LONG, # Number of rows per strip
 145                                 'StripByteCounts' => Exif::SHORT_OR_LONG, # Bytes per compressed strip
 146                                 'JPEGInterchangeFormat' => Exif::SHORT_OR_LONG, # Offset to JPEG SOI
 147                                 'JPEGInterchangeFormatLength' => Exif::SHORT_OR_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                                 # Color space transformation matrix coefficients #p27
 154                                 'YCbCrCoefficients' => array( Exif::RATIONAL, 3 ),
 155                                 'ReferenceBlackWhite' => array( Exif::RATIONAL, 6 ), # Pair of black and white reference values
 156
 157                                 # Other tags
 158                                 'DateTime' => Exif::ASCII, # File change date and time
 159                                 'ImageDescription' => Exif::ASCII, # Image title
 160                                 'Make' => Exif::ASCII, # Image input equipment manufacturer
 161                                 'Model' => Exif::ASCII, # Image input equipment model
 162                                 'Software' => Exif::ASCII, # Software used
 163                                 'Artist' => Exif::ASCII, # Person who created the image
 164                                 'Copyright' => Exif::ASCII, # Copyright holder
 165                         ),
 166
 167                         # Exif IFD Attribute Information (p30-31)
 168                         'EXIF' => array(
 169                                 # @todo NOTE: Nonexistence of this field is taken to mean nonconformance
 170                                 # to the Exif 2.1 AND 2.2 standards
 171                                 'ExifVersion' => Exif::UNDEFINED, # Exif version
 172                                 'FlashPixVersion' => Exif::UNDEFINED, # Supported Flashpix version #p32
 173
 174                                 # Tags relating to Image Data Characteristics
 175                                 'ColorSpace' => Exif::SHORT, # Color space information #p32
 176
 177                                 # Tags relating to image configuration
 178                                 'ComponentsConfiguration' => Exif::UNDEFINED, # Meaning of each component #p33
 179                                 'CompressedBitsPerPixel' => Exif::RATIONAL, # Image compression mode
 180                                 'PixelYDimension' => Exif::SHORT_OR_LONG, # Valid image width
 181                                 'PixelXDimension' => Exif::SHORT_OR_LONG, # Valid image height
 182
 183                                 # Tags relating to related user information
 184                                 'MakerNote' => Exif::IGNORE, # Manufacturer notes
 185                                 'UserComment' => Exif::UNDEFINED, # User comments #p34
 186
 187                                 # Tags relating to related file information
 188                                 'RelatedSoundFile' => Exif::ASCII, # Related audio file
 189
 190                                 # Tags relating to date and time
 191                                 'DateTimeOriginal' => Exif::ASCII, # Date and time of original data generation #p36
 192                                 'DateTimeDigitized' => Exif::ASCII, # Date and time of original data generation
 193                                 'SubSecTime' => Exif::ASCII, # DateTime subseconds
 194                                 'SubSecTimeOriginal' => Exif::ASCII, # DateTimeOriginal subseconds
 195                                 'SubSecTimeDigitized' => Exif::ASCII, # DateTimeDigitized subseconds
 196
 197                                 # Tags relating to picture-taking conditions (p31)
 198                                 'ExposureTime' => Exif::RATIONAL, # Exposure time
 199                                 'FNumber' => Exif::RATIONAL, # F Number
 200                                 'ExposureProgram' => Exif::SHORT, # Exposure Program #p38
 201                                 'SpectralSensitivity' => Exif::ASCII, # Spectral sensitivity
 202                                 'ISOSpeedRatings' => Exif::SHORT, # ISO speed rating
 203                                 'OECF' => Exif::IGNORE,
 204                                 # Optoelectronic conversion factor. Note: We don't have support for this atm.
 205                                 'ShutterSpeedValue' => Exif::SRATIONAL, # Shutter speed
 206                                 'ApertureValue' => Exif::RATIONAL, # Aperture
 207                                 'BrightnessValue' => Exif::SRATIONAL, # Brightness
 208                                 'ExposureBiasValue' => Exif::SRATIONAL, # Exposure bias
 209                                 'MaxApertureValue' => Exif::RATIONAL, # Maximum land aperture
 210                                 'SubjectDistance' => Exif::RATIONAL, # Subject distance
 211                                 'MeteringMode' => Exif::SHORT, # Metering mode #p40
 212                                 'LightSource' => Exif::SHORT, # Light source #p40-41
 213                                 'Flash' => Exif::SHORT, # Flash #p41-42
 214                                 'FocalLength' => Exif::RATIONAL, # Lens focal length
 215                                 'SubjectArea' => array( Exif::SHORT, 4 ), # Subject area
 216                                 'FlashEnergy' => Exif::RATIONAL, # Flash energy
 217                                 'SpatialFrequencyResponse' => Exif::IGNORE, # Spatial frequency response. Not supported atm.
 218                                 'FocalPlaneXResolution' => Exif::RATIONAL, # Focal plane X resolution
 219                                 'FocalPlaneYResolution' => Exif::RATIONAL, # Focal plane Y resolution
 220                                 'FocalPlaneResolutionUnit' => Exif::SHORT, # Focal plane resolution unit #p46
 221                                 'SubjectLocation' => array( Exif::SHORT, 2 ), # Subject location
 222                                 'ExposureIndex' => Exif::RATIONAL, # Exposure index
 223                                 'SensingMethod' => Exif::SHORT, # Sensing method #p46
 224                                 'FileSource' => Exif::UNDEFINED, # File source #p47
 225                                 'SceneType' => Exif::UNDEFINED, # Scene type #p47
 226                                 'CFAPattern' => Exif::IGNORE, # CFA pattern. not supported atm.
 227                                 'CustomRendered' => Exif::SHORT, # Custom image processing #p48
 228                                 'ExposureMode' => Exif::SHORT, # Exposure mode #p48
 229                                 'WhiteBalance' => Exif::SHORT, # White Balance #p49
 230                                 'DigitalZoomRatio' => Exif::RATIONAL, # Digital zoom ration
 231                                 'FocalLengthIn35mmFilm' => Exif::SHORT, # Focal length in 35 mm film
 232                                 'SceneCaptureType' => Exif::SHORT, # Scene capture type #p49
 233                                 'GainControl' => Exif::SHORT, # Scene control #p49-50
 234                                 'Contrast' => Exif::SHORT, # Contrast #p50
 235                                 'Saturation' => Exif::SHORT, # Saturation #p50
 236                                 'Sharpness' => Exif::SHORT, # Sharpness #p50
 237                                 'DeviceSettingDescription' => Exif::IGNORE,
 238                                 # Device settings description. This could maybe be supported. Need to find an
 239                                 # example file that uses this to see if it has stuff of interest in it.
 240                                 'SubjectDistanceRange' => Exif::SHORT, # Subject distance range #p51
 241
 242                                 'ImageUniqueID' => Exif::ASCII, # Unique image ID
 243                         ),
 244
 245                         # GPS Attribute Information (p52)
 246                         'GPS' => array(
 247                                 'GPSVersion' => Exif::UNDEFINED,
 248                                 # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 249                                 # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 250                                 'GPSLatitudeRef' => Exif::ASCII, # North or South Latitude #p52-53
 251                                 'GPSLatitude' => array( Exif::RATIONAL, 3 ), # Latitude
 252                                 'GPSLongitudeRef' => Exif::ASCII, # East or West Longitude #p53
 253                                 'GPSLongitude' => array( Exif::RATIONAL, 3 ), # Longitude
 254                                 'GPSAltitudeRef' => Exif::UNDEFINED,
 255                                 # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 256                                 # but php seems to disagree.
 257                                 'GPSAltitude' => Exif::RATIONAL, # Altitude
 258                                 'GPSTimeStamp' => array( Exif::RATIONAL, 3 ), # GPS time (atomic clock)
 259                                 'GPSSatellites' => Exif::ASCII, # Satellites used for measurement
 260                                 'GPSStatus' => Exif::ASCII, # Receiver status #p54
 261                                 'GPSMeasureMode' => Exif::ASCII, # Measurement mode #p54-55
 262                                 'GPSDOP' => Exif::RATIONAL, # Measurement precision
 263                                 'GPSSpeedRef' => Exif::ASCII, # Speed unit #p55
 264                                 'GPSSpeed' => Exif::RATIONAL, # Speed of GPS receiver
 265                                 'GPSTrackRef' => Exif::ASCII, # Reference for direction of movement #p55
 266                                 'GPSTrack' => Exif::RATIONAL, # Direction of movement
 267                                 'GPSImgDirectionRef' => Exif::ASCII, # Reference for direction of image #p56
 268                                 'GPSImgDirection' => Exif::RATIONAL, # Direction of image
 269                                 'GPSMapDatum' => Exif::ASCII, # Geodetic survey data used
 270                                 'GPSDestLatitudeRef' => Exif::ASCII, # Reference for latitude of destination #p56
 271                                 'GPSDestLatitude' => array( Exif::RATIONAL, 3 ), # Latitude destination
 272                                 'GPSDestLongitudeRef' => Exif::ASCII, # Reference for longitude of destination #p57
 273                                 'GPSDestLongitude' => array( Exif::RATIONAL, 3 ), # Longitude of destination
 274                                 'GPSDestBearingRef' => Exif::ASCII, # Reference for bearing of destination #p57
 275                                 'GPSDestBearing' => Exif::RATIONAL, # Bearing of destination
 276                                 'GPSDestDistanceRef' => Exif::ASCII, # Reference for distance to destination #p57-58
 277                                 'GPSDestDistance' => Exif::RATIONAL, # Distance to destination
 278                                 'GPSProcessingMethod' => Exif::UNDEFINED, # Name of GPS processing method
 279                                 'GPSAreaInformation' => Exif::UNDEFINED, # Name of GPS area
 280                                 'GPSDateStamp' => Exif::ASCII, # GPS date
 281                                 'GPSDifferential' => Exif::SHORT, # GPS differential correction
 282                         ),
 283                 );
 284
 285                 $this->file = $file;
 286                 $this->basename = wfBaseName( $this->file );
 287                 if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 288                         $this->byteOrder = $byteOrder;
 289                 } else {
 290                         // Only give a warning for b/c, since originally we didn't
 291                         // require this. The number of things affected by this is
 292                         // rather small.
 293                         wfWarn( 'Exif class did not have byte order specified. ' .
 294                                 'Some properties may be decoded incorrectly.' );
 295                         $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 296                 }
 297
 298                 $this->debugFile( $this->basename, __FUNCTION__, true );
 299                 if ( function_exists( 'exif_read_data' ) ) {
 300                         wfSuppressWarnings();
 301                         $data = exif_read_data( $this->file, 0, true );
 302                         wfRestoreWarnings();
 303                 } else {
 304                         throw new MWException( "Internal error: exif_read_data not present. " .
 305                                 "\$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 ?: 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'] )
 375                         && isset( $this->mFilteredExifData['GPSAltitudeRef'] )
 376                 ) {
 377                         // We know altitude data is a <num>/<denom> from the validation
 378                         // functions ran earlier. But multiplying such a string by -1
 379                         // doesn't work well, so convert.
 380                         list( $num, $denom ) = explode( '/', $this->mFilteredExifData['GPSAltitude'] );
 381                         $this->mFilteredExifData['GPSAltitude'] = $num / $denom;
 382
 383                         if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 384                                 $this->mFilteredExifData['GPSAltitude'] *= -1;
 385                         }
 386                         unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 387                 }
 388
 389                 $this->exifPropToOrd( 'FileSource' );
 390                 $this->exifPropToOrd( 'SceneType' );
 391
 392                 $this->charCodeString( 'UserComment' );
 393                 $this->charCodeString( 'GPSProcessingMethod' );
 394                 $this->charCodeString( 'GPSAreaInformation' );
 395
 396                 //ComponentsConfiguration should really be an array instead of a string...
 397                 //This turns a string of binary numbers into an array of numbers.
 398
 399                 if ( isset( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 400                         $val = $this->mFilteredExifData['ComponentsConfiguration'];
 401                         $ccVals = array();
 402
 403                         $strLen = strlen( $val );
 404                         for ( $i = 0; $i < $strLen; $i++ ) {
 405                                 $ccVals[$i] = ord( substr( $val, $i, 1 ) );
 406                         }
 407                         $ccVals['_type'] = 'ol'; //this is for formatting later.
 408                         $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 409                 }
 410
 411                 //GPSVersion(ID) is treated as the wrong type by php exif support.
 412                 //Go through each byte turning it into a version string.
 413                 //For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 414
 415                 //Also change exif tag name from GPSVersion (what php exif thinks it is)
 416                 //to GPSVersionID (what the exif standard thinks it is).
 417
 418                 if ( isset( $this->mFilteredExifData['GPSVersion'] ) ) {
 419                         $val = $this->mFilteredExifData['GPSVersion'];
 420                         $newVal = '';
 421
 422                         $strLen = strlen( $val );
 423                         for ( $i = 0; $i < $strLen; $i++ ) {
 424                                 if ( $i !== 0 ) {
 425                                         $newVal .= '.';
 426                                 }
 427                                 $newVal .= ord( substr( $val, $i, 1 ) );
 428                         }
 429
 430                         if ( $this->byteOrder === 'LE' ) {
 431                                 // Need to reverse the string
 432                                 $newVal2 = '';
 433                                 for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 434                                         $newVal2 .= substr( $newVal, $i, 1 );
 435                                 }
 436                                 $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 437                         } else {
 438                                 $this->mFilteredExifData['GPSVersionID'] = $newVal;
 439                         }
 440                         unset( $this->mFilteredExifData['GPSVersion'] );
 441                 }
 442         }
 443
 444         /**
 445          * Do userComment tags and similar. See pg. 34 of exif standard.
 446          * basically first 8 bytes is charset, rest is value.
 447          * This has not been tested on any shift-JIS strings.
 448          * @param string $prop prop name.
 449          */
 450         private function charCodeString( $prop ) {
 451                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 452
 453                         if ( strlen( $this->mFilteredExifData[$prop] ) <= 8 ) {
 454                                 //invalid. Must be at least 9 bytes long.
 455
 456                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, false );
 457                                 unset( $this->mFilteredExifData[$prop] );
 458
 459                                 return;
 460                         }
 461                         $charCode = substr( $this->mFilteredExifData[$prop], 0, 8 );
 462                         $val = substr( $this->mFilteredExifData[$prop], 8 );
 463
 464                         switch ( $charCode ) {
 465                                 case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 466                                         //JIS
 467                                         $charset = "Shift-JIS";
 468                                         break;
 469                                 case "UNICODE\x00":
 470                                         $charset = "UTF-16" . $this->byteOrder;
 471                                         break;
 472                                 default: //ascii or undefined.
 473                                         $charset = "";
 474                                         break;
 475                         }
 476                         // This could possibly check to see if iconv is really installed
 477                         // or if we're using the compatibility wrapper in globalFunctions.php
 478                         if ( $charset ) {
 479                                 wfSuppressWarnings();
 480                                 $val = iconv( $charset, 'UTF-8//IGNORE', $val );
 481                                 wfRestoreWarnings();
 482                         } else {
 483                                 // if valid utf-8, assume that, otherwise assume windows-1252
 484                                 $valCopy = $val;
 485                                 UtfNormal::quickIsNFCVerify( $valCopy ); //validates $valCopy.
 486                                 if ( $valCopy !== $val ) {
 487                                         wfSuppressWarnings();
 488                                         $val = iconv( 'Windows-1252', 'UTF-8//IGNORE', $val );
 489                                         wfRestoreWarnings();
 490                                 }
 491                         }
 492
 493                         //trim and check to make sure not only whitespace.
 494                         $val = trim( $val );
 495                         if ( strlen( $val ) === 0 ) {
 496                                 //only whitespace.
 497                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, "$prop: Is only whitespace" );
 498                                 unset( $this->mFilteredExifData[$prop] );
 499
 500                                 return;
 501                         }
 502
 503                         //all's good.
 504                         $this->mFilteredExifData[$prop] = $val;
 505                 }
 506         }
 507
 508         /**
 509          * Convert an Exif::UNDEFINED from a raw binary string
 510          * to its value. This is sometimes needed depending on
 511          * the type of UNDEFINED field
 512          * @param string $prop Name of property
 513          */
 514         private function exifPropToOrd( $prop ) {
 515                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 516                         $this->mFilteredExifData[$prop] = ord( $this->mFilteredExifData[$prop] );
 517                 }
 518         }
 519
 520         /**
 521          * Convert gps in exif form to a single floating point number
 522          * for example 10 degress 20`40`` S -> -10.34444
 523          * @param string $prop A GPS coordinate exif tag name (like GPSLongitude)
 524          */
 525         private function exifGPStoNumber( $prop ) {
 526                 $loc =& $this->mFilteredExifData[$prop];
 527                 $dir =& $this->mFilteredExifData[$prop . 'Ref'];
 528                 $res = false;
 529
 530                 if ( isset( $loc ) && isset( $dir )
 531                         && ( $dir === 'N' || $dir === 'S' || $dir === 'E' || $dir === 'W' )
 532                 ) {
 533                         list( $num, $denom ) = explode( '/', $loc[0] );
 534                         $res = $num / $denom;
 535                         list( $num, $denom ) = explode( '/', $loc[1] );
 536                         $res += ( $num / $denom ) * ( 1 / 60 );
 537                         list( $num, $denom ) = explode( '/', $loc[2] );
 538                         $res += ( $num / $denom ) * ( 1 / 3600 );
 539
 540                         if ( $dir === 'S' || $dir === 'W' ) {
 541                                 $res *= -1; // make negative
 542                         }
 543                 }
 544
 545                 // update the exif records.
 546
 547                 if ( $res !== false ) { // using !== as $res could potentially be 0
 548                         $this->mFilteredExifData[$prop] = $res;
 549                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 550                 } else { // if invalid
 551                         unset( $this->mFilteredExifData[$prop] );
 552                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 553                 }
 554         }
 555
 556         /**
 557          * Use FormatMetadata to create formatted values for display to user
 558          * (is this ever used?)
 559          *
 560          * @deprecated since 1.18
 561          */
 562         function makeFormattedData() {
 563                 wfDeprecated( __METHOD__, '1.18' );
 564                 $this->mFormattedExifData = FormatMetadata::getFormattedData(
 565                         $this->mFilteredExifData );
 566         }
 567
 568         /**#@-*/
 569
 570         /**#@+
 571          * @return array
 572          */
 573         /**
 574          * Get $this->mRawExifData
 575          * @return array
 576          */
 577         function getData() {
 578                 return $this->mRawExifData;
 579         }
 580
 581         /**
 582          * Get $this->mFilteredExifData
 583          */
 584         function getFilteredData() {
 585                 return $this->mFilteredExifData;
 586         }
 587
 588         /**
 589          * Get $this->mFormattedExifData
 590          *
 591          * This returns the data for display to user.
 592          * Its unclear if this is ever used.
 593          *
 594          * @deprecated since 1.18
 595          */
 596         function getFormattedData() {
 597                 wfDeprecated( __METHOD__, '1.18' );
 598                 if ( !$this->mFormattedExifData ) {
 599                         $this->makeFormattedData();
 600                 }
 601
 602                 return $this->mFormattedExifData;
 603         }
 604
 605         /**#@-*/
 606
 607         /**
 608          * The version of the output format
 609          *
 610          * Before the actual metadata information is saved in the database we
 611          * strip some of it since we don't want to save things like thumbnails
 612          * which usually accompany Exif data. This value gets saved in the
 613          * database along with the actual Exif data, and if the version in the
 614          * database doesn't equal the value returned by this function the Exif
 615          * data is regenerated.
 616          *
 617          * @return int
 618          */
 619         public static function version() {
 620                 return 2; // We don't need no bloddy constants!
 621         }
 622
 623         /**#@+
 624          * Validates if a tag value is of the type it should be according to the Exif spec
 625          *
 626          * @param mixed $in The input value to check
 627          * @return bool
 628          */
 629         private function isByte( $in ) {
 630                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 255 ) {
 631                         $this->debug( $in, __FUNCTION__, true );
 632
 633                         return true;
 634                 } else {
 635                         $this->debug( $in, __FUNCTION__, false );
 636
 637                         return false;
 638                 }
 639         }
 640
 641         /**
 642          * @param mixed $in The input value to check
 643          * @return bool
 644          */
 645         private function isASCII( $in ) {
 646                 if ( is_array( $in ) ) {
 647                         return false;
 648                 }
 649
 650                 if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 651                         $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 652
 653                         return false;
 654                 }
 655
 656                 if ( preg_match( '/^\s*$/', $in ) ) {
 657                         $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 658
 659                         return false;
 660                 }
 661
 662                 return true;
 663         }
 664
 665         /**
 666          * @param mixed $in The input value to check
 667          * @return bool
 668          */
 669         private function isShort( $in ) {
 670                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 65536 ) {
 671                         $this->debug( $in, __FUNCTION__, true );
 672
 673                         return true;
 674                 } else {
 675                         $this->debug( $in, __FUNCTION__, false );
 676
 677                         return false;
 678                 }
 679         }
 680
 681         /**
 682          * @param mixed $in The input value to check
 683          * @return bool
 684          */
 685         private function isLong( $in ) {
 686                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 4294967296 ) {
 687                         $this->debug( $in, __FUNCTION__, true );
 688
 689                         return true;
 690                 } else {
 691                         $this->debug( $in, __FUNCTION__, false );
 692
 693                         return false;
 694                 }
 695         }
 696
 697         /**
 698          * @param mixed $in The input value to check
 699          * @return bool
 700          */
 701         private function isRational( $in ) {
 702                 $m = array();
 703
 704                 # Avoid division by zero
 705                 if ( !is_array( $in )
 706                         && preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 707                 ) {
 708                         return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 709                 } else {
 710                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 711
 712                         return false;
 713                 }
 714         }
 715
 716         /**
 717          * @param mixed $in The input value to check
 718          * @return bool
 719          */
 720         private function isUndefined( $in ) {
 721                 $this->debug( $in, __FUNCTION__, true );
 722
 723                 return true;
 724         }
 725
 726         /**
 727          * @param mixed $in The input value to check
 728          * @return bool
 729          */
 730         private function isSlong( $in ) {
 731                 if ( $this->isLong( abs( $in ) ) ) {
 732                         $this->debug( $in, __FUNCTION__, true );
 733
 734                         return true;
 735                 } else {
 736                         $this->debug( $in, __FUNCTION__, false );
 737
 738                         return false;
 739                 }
 740         }
 741
 742         /**
 743          * @param mixed $in The input value to check
 744          * @return bool
 745          */
 746         private function isSrational( $in ) {
 747                 $m = array();
 748
 749                 # Avoid division by zero
 750                 if ( !is_array( $in ) &&
 751                         preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 752                 ) {
 753                         return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 754                 } else {
 755                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 756
 757                         return false;
 758                 }
 759         }
 760
 761         /**#@-*/
 762
 763         /**
 764          * Validates if a tag has a legal value according to the Exif spec
 765          *
 766          * @param string $section section where tag is located.
 767          * @param string $tag the tag to check.
 768          * @param mixed $val The value of the tag.
 769          * @param bool $recursive True if called recursively for array types.
 770          * @return bool
 771          */
 772         private function validate( $section, $tag, $val, $recursive = false ) {
 773                 $debug = "tag is '$tag'";
 774                 $etype = $this->mExifTags[$section][$tag];
 775                 $ecount = 1;
 776                 if ( is_array( $etype ) ) {
 777                         list( $etype, $ecount ) = $etype;
 778                         if ( $recursive ) {
 779                                 $ecount = 1; // checking individual elements
 780                         }
 781                 }
 782                 $count = count( $val );
 783                 if ( $ecount != $count ) {
 784                         $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 785
 786                         return false;
 787                 }
 788                 if ( $count > 1 ) {
 789                         foreach ( $val as $v ) {
 790                                 if ( !$this->validate( $section, $tag, $v, true ) ) {
 791                                         return false;
 792                                 }
 793                         }
 794
 795                         return true;
 796                 }
 797                 // Does not work if not typecast
 798                 switch ( (string)$etype ) {
 799                         case (string)Exif::BYTE:
 800                                 $this->debug( $val, __FUNCTION__, $debug );
 801
 802                                 return $this->isByte( $val );
 803                         case (string)Exif::ASCII:
 804                                 $this->debug( $val, __FUNCTION__, $debug );
 805
 806                                 return $this->isASCII( $val );
 807                         case (string)Exif::SHORT:
 808                                 $this->debug( $val, __FUNCTION__, $debug );
 809
 810                                 return $this->isShort( $val );
 811                         case (string)Exif::LONG:
 812                                 $this->debug( $val, __FUNCTION__, $debug );
 813
 814                                 return $this->isLong( $val );
 815                         case (string)Exif::RATIONAL:
 816                                 $this->debug( $val, __FUNCTION__, $debug );
 817
 818                                 return $this->isRational( $val );
 819                         case (string)Exif::SHORT_OR_LONG:
 820                                 $this->debug( $val, __FUNCTION__, $debug );
 821
 822                                 return $this->isShort( $val ) || $this->isLong( $val );
 823                         case (string)Exif::UNDEFINED:
 824                                 $this->debug( $val, __FUNCTION__, $debug );
 825
 826                                 return $this->isUndefined( $val );
 827                         case (string)Exif::SLONG:
 828                                 $this->debug( $val, __FUNCTION__, $debug );
 829
 830                                 return $this->isSlong( $val );
 831                         case (string)Exif::SRATIONAL:
 832                                 $this->debug( $val, __FUNCTION__, $debug );
 833
 834                                 return $this->isSrational( $val );
 835                         case (string)Exif::IGNORE:
 836                                 $this->debug( $val, __FUNCTION__, $debug );
 837
 838                                 return false;
 839                         default:
 840                                 $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 841
 842                                 return false;
 843                 }
 844         }
 845
 846         /**
 847          * Convenience function for debugging output
 848          *
 849          * @param mixed $in Arrays will be processed with print_r().
 850          * @param string $fname Function name to log.
 851          * @param string|bool|null $action Default null.
 852          */
 853         private function debug( $in, $fname, $action = null ) {
 854                 if ( !$this->log ) {
 855                         return;
 856                 }
 857                 $type = gettype( $in );
 858                 $class = ucfirst( __CLASS__ );
 859                 if ( is_array( $in ) ) {
 860                         $in = print_r( $in, true );
 861                 }
 862
 863                 if ( $action === true ) {
 864                         wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)\n" );
 865                 } elseif ( $action === false ) {
 866                         wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)\n" );
 867                 } elseif ( $action === null ) {
 868                         wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)\n" );
 869                 } else {
 870                         wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')\n" );
 871                 }
 872         }
 873
 874         /**
 875          * Convenience function for debugging output
 876          *
 877          * @param string $fname the name of the function calling this function
 878          * @param bool $io Specify whether we're beginning or ending
 879          */
 880         private function debugFile( $fname, $io ) {
 881                 if ( !$this->log ) {
 882                         return;
 883                 }
 884                 $class = ucfirst( __CLASS__ );
 885                 if ( $io ) {
 886                         wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'\n" );
 887                 } else {
 888                         wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'\n" );
 889                 }
 890         }
 891 }