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