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 string The file being processed */
  85         private $file;
  86
  87         /** @var string The basename of the file being processed */
  88         private $basename;
  89
  90         /** @var string The private log to log to, e.g. 'exif' */
  91         private $log = false;
  92
  93         /** @var string The byte order of the file. Needed because php's extension
  94          *    doesn't fully process some obscure props.
  95          */
  96         private $byteOrder;
  97
  98         /**
  99          * Constructor
 100          *
 101          * @param string $file Filename.
 102          * @param string $byteOrder Type of byte ordering either 'BE' (Big Endian)
 103          *   or 'LE' (Little Endian). Default ''.
 104          * @throws MWException
 105          * @todo FIXME: The following are broke:
 106          *   SubjectArea. Need to test the more obscure tags.
 107          *   DigitalZoomRatio = 0/0 is rejected. need to determine if that's valid.
 108          *   Possibly should treat 0/0 = 0. need to read exif spec on that.
 109          */
 110         function __construct( $file, $byteOrder = '' ) {
 111                 /**
 112                  * Page numbers here refer to pages in the Exif 2.2 standard
 113                  *
 114                  * Note, Exif::UNDEFINED is treated as a string, not as an array of bytes
 115                  * so don't put a count parameter for any UNDEFINED values.
 116                  *
 117                  * @link http://exif.org/Exif2-2.PDF The Exif 2.2 specification
 118                  */
 119                 $this->mExifTags = [
 120                         'COMPUTED' => [
 121                                 'Width' => Exif::SHORT_OR_LONG, # Image width
 122                                 'Height' => Exif::SHORT_OR_LONG, # Image height
 123                         ],
 124
 125                         # TIFF Rev. 6.0 Attribute Information (p22)
 126                         'IFD0' => [
 127                                 # Tags relating to image structure
 128                                 'ImageWidth' => Exif::SHORT_OR_LONG, # Image width
 129                                 'ImageLength' => Exif::SHORT_OR_LONG, # Image height
 130                                 'BitsPerSample' => [ 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' => [ 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' => [ Exif::RATIONAL, 2 ], # White point chromaticity
 154                                 'PrimaryChromaticities' => [ Exif::RATIONAL, 6 ], # Chromaticities of primarities
 155                                 # Color space transformation matrix coefficients #p27
 156                                 'YCbCrCoefficients' => [ Exif::RATIONAL, 3 ],
 157                                 'ReferenceBlackWhite' => [ Exif::RATIONAL, 6 ], # Pair of black and white reference values
 158
 159                                 # Other tags
 160                                 'DateTime' => Exif::ASCII, # File change date and time
 161                                 'ImageDescription' => Exif::ASCII, # Image title
 162                                 'Make' => Exif::ASCII, # Image input equipment manufacturer
 163                                 'Model' => Exif::ASCII, # Image input equipment model
 164                                 'Software' => Exif::ASCII, # Software used
 165                                 'Artist' => Exif::ASCII, # Person who created the image
 166                                 'Copyright' => Exif::ASCII, # Copyright holder
 167                         ],
 168
 169                         # Exif IFD Attribute Information (p30-31)
 170                         'EXIF' => [
 171                                 # @todo NOTE: Nonexistence of this field is taken to mean nonconformance
 172                                 # to the Exif 2.1 AND 2.2 standards
 173                                 'ExifVersion' => Exif::UNDEFINED, # Exif version
 174                                 'FlashPixVersion' => Exif::UNDEFINED, # Supported Flashpix version #p32
 175
 176                                 # Tags relating to Image Data Characteristics
 177                                 'ColorSpace' => Exif::SHORT, # Color space information #p32
 178
 179                                 # Tags relating to image configuration
 180                                 'ComponentsConfiguration' => Exif::UNDEFINED, # Meaning of each component #p33
 181                                 'CompressedBitsPerPixel' => Exif::RATIONAL, # Image compression mode
 182                                 'PixelYDimension' => Exif::SHORT_OR_LONG, # Valid image height
 183                                 'PixelXDimension' => Exif::SHORT_OR_LONG, # Valid image width
 184
 185                                 # Tags relating to related user information
 186                                 'MakerNote' => Exif::IGNORE, # Manufacturer notes
 187                                 'UserComment' => Exif::UNDEFINED, # User comments #p34
 188
 189                                 # Tags relating to related file information
 190                                 'RelatedSoundFile' => Exif::ASCII, # Related audio file
 191
 192                                 # Tags relating to date and time
 193                                 'DateTimeOriginal' => Exif::ASCII, # Date and time of original data generation #p36
 194                                 'DateTimeDigitized' => Exif::ASCII, # Date and time of original data generation
 195                                 'SubSecTime' => Exif::ASCII, # DateTime subseconds
 196                                 'SubSecTimeOriginal' => Exif::ASCII, # DateTimeOriginal subseconds
 197                                 'SubSecTimeDigitized' => Exif::ASCII, # DateTimeDigitized subseconds
 198
 199                                 # Tags relating to picture-taking conditions (p31)
 200                                 'ExposureTime' => Exif::RATIONAL, # Exposure time
 201                                 'FNumber' => Exif::RATIONAL, # F Number
 202                                 'ExposureProgram' => Exif::SHORT, # Exposure Program #p38
 203                                 'SpectralSensitivity' => Exif::ASCII, # Spectral sensitivity
 204                                 'ISOSpeedRatings' => Exif::SHORT, # ISO speed rating
 205                                 'OECF' => Exif::IGNORE,
 206                                 # Optoelectronic conversion factor. Note: We don't have support for this atm.
 207                                 'ShutterSpeedValue' => Exif::SRATIONAL, # Shutter speed
 208                                 'ApertureValue' => Exif::RATIONAL, # Aperture
 209                                 'BrightnessValue' => Exif::SRATIONAL, # Brightness
 210                                 'ExposureBiasValue' => Exif::SRATIONAL, # Exposure bias
 211                                 'MaxApertureValue' => Exif::RATIONAL, # Maximum land aperture
 212                                 'SubjectDistance' => Exif::RATIONAL, # Subject distance
 213                                 'MeteringMode' => Exif::SHORT, # Metering mode #p40
 214                                 'LightSource' => Exif::SHORT, # Light source #p40-41
 215                                 'Flash' => Exif::SHORT, # Flash #p41-42
 216                                 'FocalLength' => Exif::RATIONAL, # Lens focal length
 217                                 'SubjectArea' => [ Exif::SHORT, 4 ], # Subject area
 218                                 'FlashEnergy' => Exif::RATIONAL, # Flash energy
 219                                 'SpatialFrequencyResponse' => Exif::IGNORE, # Spatial frequency response. Not supported atm.
 220                                 'FocalPlaneXResolution' => Exif::RATIONAL, # Focal plane X resolution
 221                                 'FocalPlaneYResolution' => Exif::RATIONAL, # Focal plane Y resolution
 222                                 'FocalPlaneResolutionUnit' => Exif::SHORT, # Focal plane resolution unit #p46
 223                                 'SubjectLocation' => [ Exif::SHORT, 2 ], # Subject location
 224                                 'ExposureIndex' => Exif::RATIONAL, # Exposure index
 225                                 'SensingMethod' => Exif::SHORT, # Sensing method #p46
 226                                 'FileSource' => Exif::UNDEFINED, # File source #p47
 227                                 'SceneType' => Exif::UNDEFINED, # Scene type #p47
 228                                 'CFAPattern' => Exif::IGNORE, # CFA pattern. not supported atm.
 229                                 'CustomRendered' => Exif::SHORT, # Custom image processing #p48
 230                                 'ExposureMode' => Exif::SHORT, # Exposure mode #p48
 231                                 'WhiteBalance' => Exif::SHORT, # White Balance #p49
 232                                 'DigitalZoomRatio' => Exif::RATIONAL, # Digital zoom ration
 233                                 'FocalLengthIn35mmFilm' => Exif::SHORT, # Focal length in 35 mm film
 234                                 'SceneCaptureType' => Exif::SHORT, # Scene capture type #p49
 235                                 'GainControl' => Exif::SHORT, # Scene control #p49-50
 236                                 'Contrast' => Exif::SHORT, # Contrast #p50
 237                                 'Saturation' => Exif::SHORT, # Saturation #p50
 238                                 'Sharpness' => Exif::SHORT, # Sharpness #p50
 239                                 'DeviceSettingDescription' => Exif::IGNORE,
 240                                 # Device settings description. This could maybe be supported. Need to find an
 241                                 # example file that uses this to see if it has stuff of interest in it.
 242                                 'SubjectDistanceRange' => Exif::SHORT, # Subject distance range #p51
 243
 244                                 'ImageUniqueID' => Exif::ASCII, # Unique image ID
 245                         ],
 246
 247                         # GPS Attribute Information (p52)
 248                         'GPS' => [
 249                                 'GPSVersion' => Exif::UNDEFINED,
 250                                 # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 251                                 # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 252                                 'GPSLatitudeRef' => Exif::ASCII, # North or South Latitude #p52-53
 253                                 'GPSLatitude' => [ Exif::RATIONAL, 3 ], # Latitude
 254                                 'GPSLongitudeRef' => Exif::ASCII, # East or West Longitude #p53
 255                                 'GPSLongitude' => [ Exif::RATIONAL, 3 ], # Longitude
 256                                 'GPSAltitudeRef' => Exif::UNDEFINED,
 257                                 # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 258                                 # but php seems to disagree.
 259                                 'GPSAltitude' => Exif::RATIONAL, # Altitude
 260                                 'GPSTimeStamp' => [ Exif::RATIONAL, 3 ], # GPS time (atomic clock)
 261                                 'GPSSatellites' => Exif::ASCII, # Satellites used for measurement
 262                                 'GPSStatus' => Exif::ASCII, # Receiver status #p54
 263                                 'GPSMeasureMode' => Exif::ASCII, # Measurement mode #p54-55
 264                                 'GPSDOP' => Exif::RATIONAL, # Measurement precision
 265                                 'GPSSpeedRef' => Exif::ASCII, # Speed unit #p55
 266                                 'GPSSpeed' => Exif::RATIONAL, # Speed of GPS receiver
 267                                 'GPSTrackRef' => Exif::ASCII, # Reference for direction of movement #p55
 268                                 'GPSTrack' => Exif::RATIONAL, # Direction of movement
 269                                 'GPSImgDirectionRef' => Exif::ASCII, # Reference for direction of image #p56
 270                                 'GPSImgDirection' => Exif::RATIONAL, # Direction of image
 271                                 'GPSMapDatum' => Exif::ASCII, # Geodetic survey data used
 272                                 'GPSDestLatitudeRef' => Exif::ASCII, # Reference for latitude of destination #p56
 273                                 'GPSDestLatitude' => [ Exif::RATIONAL, 3 ], # Latitude destination
 274                                 'GPSDestLongitudeRef' => Exif::ASCII, # Reference for longitude of destination #p57
 275                                 'GPSDestLongitude' => [ Exif::RATIONAL, 3 ], # Longitude of destination
 276                                 'GPSDestBearingRef' => Exif::ASCII, # Reference for bearing of destination #p57
 277                                 'GPSDestBearing' => Exif::RATIONAL, # Bearing of destination
 278                                 'GPSDestDistanceRef' => Exif::ASCII, # Reference for distance to destination #p57-58
 279                                 'GPSDestDistance' => Exif::RATIONAL, # Distance to destination
 280                                 'GPSProcessingMethod' => Exif::UNDEFINED, # Name of GPS processing method
 281                                 'GPSAreaInformation' => Exif::UNDEFINED, # Name of GPS area
 282                                 'GPSDateStamp' => Exif::ASCII, # GPS date
 283                                 'GPSDifferential' => Exif::SHORT, # GPS differential correction
 284                         ],
 285                 ];
 286
 287                 $this->file = $file;
 288                 $this->basename = wfBaseName( $this->file );
 289                 if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 290                         $this->byteOrder = $byteOrder;
 291                 } else {
 292                         // Only give a warning for b/c, since originally we didn't
 293                         // require this. The number of things affected by this is
 294                         // rather small.
 295                         wfWarn( 'Exif class did not have byte order specified. ' .
 296                                 'Some properties may be decoded incorrectly.' );
 297                         $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 298                 }
 299
 300                 $this->debugFile( $this->basename, __FUNCTION__, true );
 301                 if ( function_exists( 'exif_read_data' ) ) {
 302                         MediaWiki\suppressWarnings();
 303                         $data = exif_read_data( $this->file, 0, true );
 304                         MediaWiki\restoreWarnings();
 305                 } else {
 306                         throw new MWException( "Internal error: exif_read_data not present. " .
 307                                 "\$wgShowEXIF may be incorrectly set or not checked by an extension." );
 308                 }
 309                 /**
 310                  * exif_read_data() will return false on invalid input, such as
 311                  * when somebody uploads a file called something.jpeg
 312                  * containing random gibberish.
 313                  */
 314                 $this->mRawExifData = $data ?: [];
 315                 $this->makeFilteredData();
 316                 $this->collapseData();
 317                 $this->debugFile( __FUNCTION__, false );
 318         }
 319
 320         /**
 321          * Make $this->mFilteredExifData
 322          */
 323         function makeFilteredData() {
 324                 $this->mFilteredExifData = [];
 325
 326                 foreach ( array_keys( $this->mRawExifData ) as $section ) {
 327                         if ( !array_key_exists( $section, $this->mExifTags ) ) {
 328                                 $this->debug( $section, __FUNCTION__, "'$section' is not a valid Exif section" );
 329                                 continue;
 330                         }
 331
 332                         foreach ( array_keys( $this->mRawExifData[$section] ) as $tag ) {
 333                                 if ( !array_key_exists( $tag, $this->mExifTags[$section] ) ) {
 334                                         $this->debug( $tag, __FUNCTION__, "'$tag' is not a valid tag in '$section'" );
 335                                         continue;
 336                                 }
 337
 338                                 $this->mFilteredExifData[$tag] = $this->mRawExifData[$section][$tag];
 339                                 // This is ok, as the tags in the different sections do not conflict.
 340                                 // except in computed and thumbnail section, which we don't use.
 341
 342                                 $value = $this->mRawExifData[$section][$tag];
 343                                 if ( !$this->validate( $section, $tag, $value ) ) {
 344                                         $this->debug( $value, __FUNCTION__, "'$tag' contained invalid data" );
 345                                         unset( $this->mFilteredExifData[$tag] );
 346                                 }
 347                         }
 348                 }
 349         }
 350
 351         /**
 352          * Collapse some fields together.
 353          * This converts some fields from exif form, to a more friendly form.
 354          * For example GPS latitude to a single number.
 355          *
 356          * The rationale behind this is that we're storing data, not presenting to the user
 357          * For example a longitude is a single number describing how far away you are from
 358          * the prime meridian. Well it might be nice to split it up into minutes and seconds
 359          * for the user, it doesn't really make sense to split a single number into 4 parts
 360          * for storage. (degrees, minutes, second, direction vs single floating point number).
 361          *
 362          * Other things this might do (not really sure if they make sense or not):
 363          * Dates -> mediawiki date format.
 364          * convert values that can be in different units to be in one standardized unit.
 365          *
 366          * As an alternative approach, some of this could be done in the validate phase
 367          * if we make up our own types like Exif::DATE.
 368          */
 369         function collapseData() {
 370
 371                 $this->exifGPStoNumber( 'GPSLatitude' );
 372                 $this->exifGPStoNumber( 'GPSDestLatitude' );
 373                 $this->exifGPStoNumber( 'GPSLongitude' );
 374                 $this->exifGPStoNumber( 'GPSDestLongitude' );
 375
 376                 if ( isset( $this->mFilteredExifData['GPSAltitude'] )
 377                         && isset( $this->mFilteredExifData['GPSAltitudeRef'] )
 378                 ) {
 379                         // We know altitude data is a <num>/<denom> from the validation
 380                         // functions ran earlier. But multiplying such a string by -1
 381                         // doesn't work well, so convert.
 382                         list( $num, $denom ) = explode( '/', $this->mFilteredExifData['GPSAltitude'] );
 383                         $this->mFilteredExifData['GPSAltitude'] = $num / $denom;
 384
 385                         if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 386                                 $this->mFilteredExifData['GPSAltitude'] *= -1;
 387                         }
 388                         unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 389                 }
 390
 391                 $this->exifPropToOrd( 'FileSource' );
 392                 $this->exifPropToOrd( 'SceneType' );
 393
 394                 $this->charCodeString( 'UserComment' );
 395                 $this->charCodeString( 'GPSProcessingMethod' );
 396                 $this->charCodeString( 'GPSAreaInformation' );
 397
 398                 // ComponentsConfiguration should really be an array instead of a string...
 399                 // This turns a string of binary numbers into an array of numbers.
 400
 401                 if ( isset( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 402                         $val = $this->mFilteredExifData['ComponentsConfiguration'];
 403                         $ccVals = [];
 404
 405                         $strLen = strlen( $val );
 406                         for ( $i = 0; $i < $strLen; $i++ ) {
 407                                 $ccVals[$i] = ord( substr( $val, $i, 1 ) );
 408                         }
 409                         $ccVals['_type'] = 'ol'; // this is for formatting later.
 410                         $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 411                 }
 412
 413                 // GPSVersion(ID) is treated as the wrong type by php exif support.
 414                 // Go through each byte turning it into a version string.
 415                 // For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 416
 417                 // Also change exif tag name from GPSVersion (what php exif thinks it is)
 418                 // to GPSVersionID (what the exif standard thinks it is).
 419
 420                 if ( isset( $this->mFilteredExifData['GPSVersion'] ) ) {
 421                         $val = $this->mFilteredExifData['GPSVersion'];
 422                         $newVal = '';
 423
 424                         $strLen = strlen( $val );
 425                         for ( $i = 0; $i < $strLen; $i++ ) {
 426                                 if ( $i !== 0 ) {
 427                                         $newVal .= '.';
 428                                 }
 429                                 $newVal .= ord( substr( $val, $i, 1 ) );
 430                         }
 431
 432                         if ( $this->byteOrder === 'LE' ) {
 433                                 // Need to reverse the string
 434                                 $newVal2 = '';
 435                                 for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 436                                         $newVal2 .= substr( $newVal, $i, 1 );
 437                                 }
 438                                 $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 439                         } else {
 440                                 $this->mFilteredExifData['GPSVersionID'] = $newVal;
 441                         }
 442                         unset( $this->mFilteredExifData['GPSVersion'] );
 443                 }
 444         }
 445
 446         /**
 447          * Do userComment tags and similar. See pg. 34 of exif standard.
 448          * basically first 8 bytes is charset, rest is value.
 449          * This has not been tested on any shift-JIS strings.
 450          * @param string $prop Prop name
 451          */
 452         private function charCodeString( $prop ) {
 453                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 454
 455                         if ( strlen( $this->mFilteredExifData[$prop] ) <= 8 ) {
 456                                 // invalid. Must be at least 9 bytes long.
 457
 458                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, false );
 459                                 unset( $this->mFilteredExifData[$prop] );
 460
 461                                 return;
 462                         }
 463                         $charCode = substr( $this->mFilteredExifData[$prop], 0, 8 );
 464                         $val = substr( $this->mFilteredExifData[$prop], 8 );
 465
 466                         switch ( $charCode ) {
 467                                 case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 468                                         // JIS
 469                                         $charset = "Shift-JIS";
 470                                         break;
 471                                 case "UNICODE\x00":
 472                                         $charset = "UTF-16" . $this->byteOrder;
 473                                         break;
 474                                 default: // ascii or undefined.
 475                                         $charset = "";
 476                                         break;
 477                         }
 478                         if ( $charset ) {
 479                                 MediaWiki\suppressWarnings();
 480                                 $val = iconv( $charset, 'UTF-8//IGNORE', $val );
 481                                 MediaWiki\restoreWarnings();
 482                         } else {
 483                                 // if valid utf-8, assume that, otherwise assume windows-1252
 484                                 $valCopy = $val;
 485                                 UtfNormal\Validator::quickIsNFCVerify( $valCopy ); // validates $valCopy.
 486                                 if ( $valCopy !== $val ) {
 487                                         MediaWiki\suppressWarnings();
 488                                         $val = iconv( 'Windows-1252', 'UTF-8//IGNORE', $val );
 489                                         MediaWiki\restoreWarnings();
 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
 558         /**#@+
 559          * @return array
 560          */
 561         /**
 562          * Get $this->mRawExifData
 563          * @return array
 564          */
 565         function getData() {
 566                 return $this->mRawExifData;
 567         }
 568
 569         /**
 570          * Get $this->mFilteredExifData
 571          * @return array
 572          */
 573         function getFilteredData() {
 574                 return $this->mFilteredExifData;
 575         }
 576
 577         /**#@-*/
 578
 579         /**
 580          * The version of the output format
 581          *
 582          * Before the actual metadata information is saved in the database we
 583          * strip some of it since we don't want to save things like thumbnails
 584          * which usually accompany Exif data. This value gets saved in the
 585          * database along with the actual Exif data, and if the version in the
 586          * database doesn't equal the value returned by this function the Exif
 587          * data is regenerated.
 588          *
 589          * @return int
 590          */
 591         public static function version() {
 592                 return 2; // We don't need no bloddy constants!
 593         }
 594
 595         /**
 596          * Validates if a tag value is of the type it should be according to the Exif spec
 597          *
 598          * @param mixed $in The input value to check
 599          * @return bool
 600          */
 601         private function isByte( $in ) {
 602                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 255 ) {
 603                         $this->debug( $in, __FUNCTION__, true );
 604
 605                         return true;
 606                 } else {
 607                         $this->debug( $in, __FUNCTION__, false );
 608
 609                         return false;
 610                 }
 611         }
 612
 613         /**
 614          * @param mixed $in The input value to check
 615          * @return bool
 616          */
 617         private function isASCII( $in ) {
 618                 if ( is_array( $in ) ) {
 619                         return false;
 620                 }
 621
 622                 if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 623                         $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 624
 625                         return false;
 626                 }
 627
 628                 if ( preg_match( '/^\s*$/', $in ) ) {
 629                         $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 630
 631                         return false;
 632                 }
 633
 634                 return true;
 635         }
 636
 637         /**
 638          * @param mixed $in The input value to check
 639          * @return bool
 640          */
 641         private function isShort( $in ) {
 642                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 65536 ) {
 643                         $this->debug( $in, __FUNCTION__, true );
 644
 645                         return true;
 646                 } else {
 647                         $this->debug( $in, __FUNCTION__, false );
 648
 649                         return false;
 650                 }
 651         }
 652
 653         /**
 654          * @param mixed $in The input value to check
 655          * @return bool
 656          */
 657         private function isLong( $in ) {
 658                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 4294967296 ) {
 659                         $this->debug( $in, __FUNCTION__, true );
 660
 661                         return true;
 662                 } else {
 663                         $this->debug( $in, __FUNCTION__, false );
 664
 665                         return false;
 666                 }
 667         }
 668
 669         /**
 670          * @param mixed $in The input value to check
 671          * @return bool
 672          */
 673         private function isRational( $in ) {
 674                 $m = [];
 675
 676                 # Avoid division by zero
 677                 if ( !is_array( $in )
 678                         && preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 679                 ) {
 680                         return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 681                 } else {
 682                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 683
 684                         return false;
 685                 }
 686         }
 687
 688         /**
 689          * @param mixed $in The input value to check
 690          * @return bool
 691          */
 692         private function isUndefined( $in ) {
 693                 $this->debug( $in, __FUNCTION__, true );
 694
 695                 return true;
 696         }
 697
 698         /**
 699          * @param mixed $in The input value to check
 700          * @return bool
 701          */
 702         private function isSlong( $in ) {
 703                 if ( $this->isLong( abs( $in ) ) ) {
 704                         $this->debug( $in, __FUNCTION__, true );
 705
 706                         return true;
 707                 } else {
 708                         $this->debug( $in, __FUNCTION__, false );
 709
 710                         return false;
 711                 }
 712         }
 713
 714         /**
 715          * @param mixed $in The input value to check
 716          * @return bool
 717          */
 718         private function isSrational( $in ) {
 719                 $m = [];
 720
 721                 # Avoid division by zero
 722                 if ( !is_array( $in ) &&
 723                         preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 724                 ) {
 725                         return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 726                 } else {
 727                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 728
 729                         return false;
 730                 }
 731         }
 732
 733         /**#@-*/
 734
 735         /**
 736          * Validates if a tag has a legal value according to the Exif spec
 737          *
 738          * @param string $section Section where tag is located.
 739          * @param string $tag The tag to check.
 740          * @param mixed $val The value of the tag.
 741          * @param bool $recursive True if called recursively for array types.
 742          * @return bool
 743          */
 744         private function validate( $section, $tag, $val, $recursive = false ) {
 745                 $debug = "tag is '$tag'";
 746                 $etype = $this->mExifTags[$section][$tag];
 747                 $ecount = 1;
 748                 if ( is_array( $etype ) ) {
 749                         list( $etype, $ecount ) = $etype;
 750                         if ( $recursive ) {
 751                                 $ecount = 1; // checking individual elements
 752                         }
 753                 }
 754                 $count = count( $val );
 755                 if ( $ecount != $count ) {
 756                         $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 757
 758                         return false;
 759                 }
 760                 if ( $count > 1 ) {
 761                         foreach ( $val as $v ) {
 762                                 if ( !$this->validate( $section, $tag, $v, true ) ) {
 763                                         return false;
 764                                 }
 765                         }
 766
 767                         return true;
 768                 }
 769                 // Does not work if not typecast
 770                 switch ( (string)$etype ) {
 771                         case (string)Exif::BYTE:
 772                                 $this->debug( $val, __FUNCTION__, $debug );
 773
 774                                 return $this->isByte( $val );
 775                         case (string)Exif::ASCII:
 776                                 $this->debug( $val, __FUNCTION__, $debug );
 777
 778                                 return $this->isASCII( $val );
 779                         case (string)Exif::SHORT:
 780                                 $this->debug( $val, __FUNCTION__, $debug );
 781
 782                                 return $this->isShort( $val );
 783                         case (string)Exif::LONG:
 784                                 $this->debug( $val, __FUNCTION__, $debug );
 785
 786                                 return $this->isLong( $val );
 787                         case (string)Exif::RATIONAL:
 788                                 $this->debug( $val, __FUNCTION__, $debug );
 789
 790                                 return $this->isRational( $val );
 791                         case (string)Exif::SHORT_OR_LONG:
 792                                 $this->debug( $val, __FUNCTION__, $debug );
 793
 794                                 return $this->isShort( $val ) || $this->isLong( $val );
 795                         case (string)Exif::UNDEFINED:
 796                                 $this->debug( $val, __FUNCTION__, $debug );
 797
 798                                 return $this->isUndefined( $val );
 799                         case (string)Exif::SLONG:
 800                                 $this->debug( $val, __FUNCTION__, $debug );
 801
 802                                 return $this->isSlong( $val );
 803                         case (string)Exif::SRATIONAL:
 804                                 $this->debug( $val, __FUNCTION__, $debug );
 805
 806                                 return $this->isSrational( $val );
 807                         case (string)Exif::IGNORE:
 808                                 $this->debug( $val, __FUNCTION__, $debug );
 809
 810                                 return false;
 811                         default:
 812                                 $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 813
 814                                 return false;
 815                 }
 816         }
 817
 818         /**
 819          * Convenience function for debugging output
 820          *
 821          * @param mixed $in Arrays will be processed with print_r().
 822          * @param string $fname Function name to log.
 823          * @param string|bool|null $action Default null.
 824          */
 825         private function debug( $in, $fname, $action = null ) {
 826                 if ( !$this->log ) {
 827                         return;
 828                 }
 829                 $type = gettype( $in );
 830                 $class = ucfirst( __CLASS__ );
 831                 if ( is_array( $in ) ) {
 832                         $in = print_r( $in, true );
 833                 }
 834
 835                 if ( $action === true ) {
 836                         wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)" );
 837                 } elseif ( $action === false ) {
 838                         wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)" );
 839                 } elseif ( $action === null ) {
 840                         wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)" );
 841                 } else {
 842                         wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')" );
 843                 }
 844         }
 845
 846         /**
 847          * Convenience function for debugging output
 848          *
 849          * @param string $fname The name of the function calling this function
 850          * @param bool $io Specify whether we're beginning or ending
 851          */
 852         private function debugFile( $fname, $io ) {
 853                 if ( !$this->log ) {
 854                         return;
 855                 }
 856                 $class = ucfirst( __CLASS__ );
 857                 if ( $io ) {
 858                         wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'" );
 859                 } else {
 860                         wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'" );
 861                 }
 862         }
 863 }