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