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         //@{
  70         /* @var array
  71          * @private
  72          */
  73
  74         /**
  75          * Exif tags grouped by category, the tagname itself is the key and the type
  76          * is the value, in the case of more than one possible value type they are
  77          * separated by commas.
  78          */
  79         var $mExifTags;
  80
  81         /**
  82          * The raw Exif data returned by exif_read_data()
  83          */
  84         var $mRawExifData;
  85
  86         /**
  87          * A Filtered version of $mRawExifData that has been pruned of invalid
  88          * tags and tags that contain content they shouldn't contain according
  89          * to the Exif specification
  90          */
  91         var $mFilteredExifData;
  92
  93         /**
  94          * Filtered and formatted Exif data, see FormatMetadata::getFormattedData()
  95          */
  96         var $mFormattedExifData;
  97
  98         //@}
  99
 100         //@{
 101         /* @var string
 102          * @private
 103          */
 104
 105         /**
 106          * The file being processed
 107          */
 108         var $file;
 109
 110         /**
 111          * The basename of the file being processed
 112          */
 113         var $basename;
 114
 115         /**
 116          * The private log to log to, e.g. 'exif'
 117          */
 118         var $log = false;
 119
 120         /**
 121          * The byte order of the file. Needed because php's
 122          * extension doesn't fully process some obscure props.
 123          */
 124         private $byteOrder;
 125
 126         //@}
 127
 128         /**
 129          * Constructor
 130          *
 131          * @param string $file filename.
 132          * @param string $byteOrder Type of byte ordering either 'BE' (Big Endian)
 133          *   or 'LE' (Little Endian). Default ''.
 134          * @throws MWException
 135          * @todo FIXME: The following are broke:
 136          * SubjectArea. Need to test the more obscure tags.
 137          *
 138          * DigitalZoomRatio = 0/0 is rejected. need to determine if that's valid.
 139          * possibly should treat 0/0 = 0. need to read exif spec on that.
 140          */
 141         function __construct( $file, $byteOrder = '' ) {
 142                 /**
 143                  * Page numbers here refer to pages in the Exif 2.2 standard
 144                  *
 145                  * Note, Exif::UNDEFINED is treated as a string, not as an array of bytes
 146                  * so don't put a count parameter for any UNDEFINED values.
 147                  *
 148                  * @link http://exif.org/Exif2-2.PDF The Exif 2.2 specification
 149                  */
 150                 $this->mExifTags = array(
 151                         # TIFF Rev. 6.0 Attribute Information (p22)
 152                         'IFD0' => array(
 153                                 # Tags relating to image structure
 154                                 'ImageWidth' => Exif::SHORT_OR_LONG, # Image width
 155                                 'ImageLength' => Exif::SHORT_OR_LONG, # Image height
 156                                 'BitsPerSample' => array( Exif::SHORT, 3 ), # Number of bits per component
 157                                 # "When a primary image is JPEG compressed, this designation is not"
 158                                 # "necessary and is omitted." (p23)
 159                                 'Compression' => Exif::SHORT, # Compression scheme #p23
 160                                 'PhotometricInterpretation' => Exif::SHORT, # Pixel composition #p23
 161                                 'Orientation' => Exif::SHORT, # Orientation of image #p24
 162                                 'SamplesPerPixel' => Exif::SHORT, # Number of components
 163                                 'PlanarConfiguration' => Exif::SHORT, # Image data arrangement #p24
 164                                 'YCbCrSubSampling' => array( Exif::SHORT, 2 ), # Subsampling ratio of Y to C #p24
 165                                 'YCbCrPositioning' => Exif::SHORT, # Y and C positioning #p24-25
 166                                 'XResolution' => Exif::RATIONAL, # Image resolution in width direction
 167                                 'YResolution' => Exif::RATIONAL, # Image resolution in height direction
 168                                 'ResolutionUnit' => Exif::SHORT, # Unit of X and Y resolution #(p26)
 169
 170                                 # Tags relating to recording offset
 171                                 'StripOffsets' => Exif::SHORT_OR_LONG, # Image data location
 172                                 'RowsPerStrip' => Exif::SHORT_OR_LONG, # Number of rows per strip
 173                                 'StripByteCounts' => Exif::SHORT_OR_LONG, # Bytes per compressed strip
 174                                 'JPEGInterchangeFormat' => Exif::SHORT_OR_LONG, # Offset to JPEG SOI
 175                                 'JPEGInterchangeFormatLength' => Exif::SHORT_OR_LONG, # Bytes of JPEG data
 176
 177                                 # Tags relating to image data characteristics
 178                                 'TransferFunction' => Exif::IGNORE, # Transfer function
 179                                 'WhitePoint' => array( Exif::RATIONAL, 2 ), # White point chromaticity
 180                                 'PrimaryChromaticities' => array( Exif::RATIONAL, 6 ), # Chromaticities of primarities
 181                                 # Color space transformation matrix coefficients #p27
 182                                 'YCbCrCoefficients' => array( Exif::RATIONAL, 3 ),
 183                                 'ReferenceBlackWhite' => array( Exif::RATIONAL, 6 ), # Pair of black and white reference values
 184
 185                                 # Other tags
 186                                 'DateTime' => Exif::ASCII, # File change date and time
 187                                 'ImageDescription' => Exif::ASCII, # Image title
 188                                 'Make' => Exif::ASCII, # Image input equipment manufacturer
 189                                 'Model' => Exif::ASCII, # Image input equipment model
 190                                 'Software' => Exif::ASCII, # Software used
 191                                 'Artist' => Exif::ASCII, # Person who created the image
 192                                 'Copyright' => Exif::ASCII, # Copyright holder
 193                         ),
 194
 195                         # Exif IFD Attribute Information (p30-31)
 196                         'EXIF' => array(
 197                                 # TODO: NOTE: Nonexistence of this field is taken to mean nonconformance
 198                                 # to the Exif 2.1 AND 2.2 standards
 199                                 'ExifVersion' => Exif::UNDEFINED, # Exif version
 200                                 'FlashPixVersion' => Exif::UNDEFINED, # Supported Flashpix version #p32
 201
 202                                 # Tags relating to Image Data Characteristics
 203                                 'ColorSpace' => Exif::SHORT, # Color space information #p32
 204
 205                                 # Tags relating to image configuration
 206                                 'ComponentsConfiguration' => Exif::UNDEFINED, # Meaning of each component #p33
 207                                 'CompressedBitsPerPixel' => Exif::RATIONAL, # Image compression mode
 208                                 'PixelYDimension' => Exif::SHORT_OR_LONG, # Valid image width
 209                                 'PixelXDimension' => Exif::SHORT_OR_LONG, # Valid image height
 210
 211                                 # Tags relating to related user information
 212                                 'MakerNote' => Exif::IGNORE, # Manufacturer notes
 213                                 'UserComment' => Exif::UNDEFINED, # User comments #p34
 214
 215                                 # Tags relating to related file information
 216                                 'RelatedSoundFile' => Exif::ASCII, # Related audio file
 217
 218                                 # Tags relating to date and time
 219                                 'DateTimeOriginal' => Exif::ASCII, # Date and time of original data generation #p36
 220                                 'DateTimeDigitized' => Exif::ASCII, # Date and time of original data generation
 221                                 'SubSecTime' => Exif::ASCII, # DateTime subseconds
 222                                 'SubSecTimeOriginal' => Exif::ASCII, # DateTimeOriginal subseconds
 223                                 'SubSecTimeDigitized' => Exif::ASCII, # DateTimeDigitized subseconds
 224
 225                                 # Tags relating to picture-taking conditions (p31)
 226                                 'ExposureTime' => Exif::RATIONAL, # Exposure time
 227                                 'FNumber' => Exif::RATIONAL, # F Number
 228                                 'ExposureProgram' => Exif::SHORT, # Exposure Program #p38
 229                                 'SpectralSensitivity' => Exif::ASCII, # Spectral sensitivity
 230                                 'ISOSpeedRatings' => Exif::SHORT, # ISO speed rating
 231                                 'OECF' => Exif::IGNORE,
 232                                 # Optoelectronic conversion factor. Note: We don't have support for this atm.
 233                                 'ShutterSpeedValue' => Exif::SRATIONAL, # Shutter speed
 234                                 'ApertureValue' => Exif::RATIONAL, # Aperture
 235                                 'BrightnessValue' => Exif::SRATIONAL, # Brightness
 236                                 'ExposureBiasValue' => Exif::SRATIONAL, # Exposure bias
 237                                 'MaxApertureValue' => Exif::RATIONAL, # Maximum land aperture
 238                                 'SubjectDistance' => Exif::RATIONAL, # Subject distance
 239                                 'MeteringMode' => Exif::SHORT, # Metering mode #p40
 240                                 'LightSource' => Exif::SHORT, # Light source #p40-41
 241                                 'Flash' => Exif::SHORT, # Flash #p41-42
 242                                 'FocalLength' => Exif::RATIONAL, # Lens focal length
 243                                 'SubjectArea' => array( Exif::SHORT, 4 ), # Subject area
 244                                 'FlashEnergy' => Exif::RATIONAL, # Flash energy
 245                                 'SpatialFrequencyResponse' => Exif::IGNORE, # Spatial frequency response. Not supported atm.
 246                                 'FocalPlaneXResolution' => Exif::RATIONAL, # Focal plane X resolution
 247                                 'FocalPlaneYResolution' => Exif::RATIONAL, # Focal plane Y resolution
 248                                 'FocalPlaneResolutionUnit' => Exif::SHORT, # Focal plane resolution unit #p46
 249                                 'SubjectLocation' => array( Exif::SHORT, 2 ), # Subject location
 250                                 'ExposureIndex' => Exif::RATIONAL, # Exposure index
 251                                 'SensingMethod' => Exif::SHORT, # Sensing method #p46
 252                                 'FileSource' => Exif::UNDEFINED, # File source #p47
 253                                 'SceneType' => Exif::UNDEFINED, # Scene type #p47
 254                                 'CFAPattern' => Exif::IGNORE, # CFA pattern. not supported atm.
 255                                 'CustomRendered' => Exif::SHORT, # Custom image processing #p48
 256                                 'ExposureMode' => Exif::SHORT, # Exposure mode #p48
 257                                 'WhiteBalance' => Exif::SHORT, # White Balance #p49
 258                                 'DigitalZoomRatio' => Exif::RATIONAL, # Digital zoom ration
 259                                 'FocalLengthIn35mmFilm' => Exif::SHORT, # Focal length in 35 mm film
 260                                 'SceneCaptureType' => Exif::SHORT, # Scene capture type #p49
 261                                 'GainControl' => Exif::SHORT, # Scene control #p49-50
 262                                 'Contrast' => Exif::SHORT, # Contrast #p50
 263                                 'Saturation' => Exif::SHORT, # Saturation #p50
 264                                 'Sharpness' => Exif::SHORT, # Sharpness #p50
 265                                 'DeviceSettingDescription' => Exif::IGNORE,
 266                                 # Device settings description. This could maybe be supported. Need to find an
 267                                 # example file that uses this to see if it has stuff of interest in it.
 268                                 'SubjectDistanceRange' => Exif::SHORT, # Subject distance range #p51
 269
 270                                 'ImageUniqueID' => Exif::ASCII, # Unique image ID
 271                         ),
 272
 273                         # GPS Attribute Information (p52)
 274                         'GPS' => array(
 275                                 'GPSVersion' => Exif::UNDEFINED,
 276                                 # Should be an array of 4 Exif::BYTE's. However php treats it as an undefined
 277                                 # Note exif standard calls this GPSVersionID, but php doesn't like the id suffix
 278                                 'GPSLatitudeRef' => Exif::ASCII, # North or South Latitude #p52-53
 279                                 'GPSLatitude' => array( Exif::RATIONAL, 3 ), # Latitude
 280                                 'GPSLongitudeRef' => Exif::ASCII, # East or West Longitude #p53
 281                                 'GPSLongitude' => array( Exif::RATIONAL, 3 ), # Longitude
 282                                 'GPSAltitudeRef' => Exif::UNDEFINED,
 283                                 # Altitude reference. Note, the exif standard says this should be an EXIF::Byte,
 284                                 # but php seems to disagree.
 285                                 'GPSAltitude' => Exif::RATIONAL, # Altitude
 286                                 'GPSTimeStamp' => array( Exif::RATIONAL, 3 ), # GPS time (atomic clock)
 287                                 'GPSSatellites' => Exif::ASCII, # Satellites used for measurement
 288                                 'GPSStatus' => Exif::ASCII, # Receiver status #p54
 289                                 'GPSMeasureMode' => Exif::ASCII, # Measurement mode #p54-55
 290                                 'GPSDOP' => Exif::RATIONAL, # Measurement precision
 291                                 'GPSSpeedRef' => Exif::ASCII, # Speed unit #p55
 292                                 'GPSSpeed' => Exif::RATIONAL, # Speed of GPS receiver
 293                                 'GPSTrackRef' => Exif::ASCII, # Reference for direction of movement #p55
 294                                 'GPSTrack' => Exif::RATIONAL, # Direction of movement
 295                                 'GPSImgDirectionRef' => Exif::ASCII, # Reference for direction of image #p56
 296                                 'GPSImgDirection' => Exif::RATIONAL, # Direction of image
 297                                 'GPSMapDatum' => Exif::ASCII, # Geodetic survey data used
 298                                 'GPSDestLatitudeRef' => Exif::ASCII, # Reference for latitude of destination #p56
 299                                 'GPSDestLatitude' => array( Exif::RATIONAL, 3 ), # Latitude destination
 300                                 'GPSDestLongitudeRef' => Exif::ASCII, # Reference for longitude of destination #p57
 301                                 'GPSDestLongitude' => array( Exif::RATIONAL, 3 ), # Longitude of destination
 302                                 'GPSDestBearingRef' => Exif::ASCII, # Reference for bearing of destination #p57
 303                                 'GPSDestBearing' => Exif::RATIONAL, # Bearing of destination
 304                                 'GPSDestDistanceRef' => Exif::ASCII, # Reference for distance to destination #p57-58
 305                                 'GPSDestDistance' => Exif::RATIONAL, # Distance to destination
 306                                 'GPSProcessingMethod' => Exif::UNDEFINED, # Name of GPS processing method
 307                                 'GPSAreaInformation' => Exif::UNDEFINED, # Name of GPS area
 308                                 'GPSDateStamp' => Exif::ASCII, # GPS date
 309                                 'GPSDifferential' => Exif::SHORT, # GPS differential correction
 310                         ),
 311                 );
 312
 313                 $this->file = $file;
 314                 $this->basename = wfBaseName( $this->file );
 315                 if ( $byteOrder === 'BE' || $byteOrder === 'LE' ) {
 316                         $this->byteOrder = $byteOrder;
 317                 } else {
 318                         // Only give a warning for b/c, since originally we didn't
 319                         // require this. The number of things affected by this is
 320                         // rather small.
 321                         wfWarn( 'Exif class did not have byte order specified. ' .
 322                                 'Some properties may be decoded incorrectly.' );
 323                         $this->byteOrder = 'BE'; // BE seems about twice as popular as LE in jpg's.
 324                 }
 325
 326                 $this->debugFile( $this->basename, __FUNCTION__, true );
 327                 if ( function_exists( 'exif_read_data' ) ) {
 328                         wfSuppressWarnings();
 329                         $data = exif_read_data( $this->file, 0, true );
 330                         wfRestoreWarnings();
 331                 } else {
 332                         throw new MWException( "Internal error: exif_read_data not present. " .
 333                                 "\$wgShowEXIF may be incorrectly set or not checked by an extension." );
 334                 }
 335                 /**
 336                  * exif_read_data() will return false on invalid input, such as
 337                  * when somebody uploads a file called something.jpeg
 338                  * containing random gibberish.
 339                  */
 340                 $this->mRawExifData = $data ?: array();
 341                 $this->makeFilteredData();
 342                 $this->collapseData();
 343                 $this->debugFile( __FUNCTION__, false );
 344         }
 345
 346         /**
 347          * Make $this->mFilteredExifData
 348          */
 349         function makeFilteredData() {
 350                 $this->mFilteredExifData = array();
 351
 352                 foreach ( array_keys( $this->mRawExifData ) as $section ) {
 353                         if ( !in_array( $section, array_keys( $this->mExifTags ) ) ) {
 354                                 $this->debug( $section, __FUNCTION__, "'$section' is not a valid Exif section" );
 355                                 continue;
 356                         }
 357
 358                         foreach ( array_keys( $this->mRawExifData[$section] ) as $tag ) {
 359                                 if ( !in_array( $tag, array_keys( $this->mExifTags[$section] ) ) ) {
 360                                         $this->debug( $tag, __FUNCTION__, "'$tag' is not a valid tag in '$section'" );
 361                                         continue;
 362                                 }
 363
 364                                 $this->mFilteredExifData[$tag] = $this->mRawExifData[$section][$tag];
 365                                 // This is ok, as the tags in the different sections do not conflict.
 366                                 // except in computed and thumbnail section, which we don't use.
 367
 368                                 $value = $this->mRawExifData[$section][$tag];
 369                                 if ( !$this->validate( $section, $tag, $value ) ) {
 370                                         $this->debug( $value, __FUNCTION__, "'$tag' contained invalid data" );
 371                                         unset( $this->mFilteredExifData[$tag] );
 372                                 }
 373                         }
 374                 }
 375         }
 376
 377         /**
 378          * Collapse some fields together.
 379          * This converts some fields from exif form, to a more friendly form.
 380          * For example GPS latitude to a single number.
 381          *
 382          * The rationale behind this is that we're storing data, not presenting to the user
 383          * For example a longitude is a single number describing how far away you are from
 384          * the prime meridian. Well it might be nice to split it up into minutes and seconds
 385          * for the user, it doesn't really make sense to split a single number into 4 parts
 386          * for storage. (degrees, minutes, second, direction vs single floating point number).
 387          *
 388          * Other things this might do (not really sure if they make sense or not):
 389          * Dates -> mediawiki date format.
 390          * convert values that can be in different units to be in one standardized unit.
 391          *
 392          * As an alternative approach, some of this could be done in the validate phase
 393          * if we make up our own types like Exif::DATE.
 394          */
 395         function collapseData() {
 396
 397                 $this->exifGPStoNumber( 'GPSLatitude' );
 398                 $this->exifGPStoNumber( 'GPSDestLatitude' );
 399                 $this->exifGPStoNumber( 'GPSLongitude' );
 400                 $this->exifGPStoNumber( 'GPSDestLongitude' );
 401
 402                 if ( isset( $this->mFilteredExifData['GPSAltitude'] )
 403                         && isset( $this->mFilteredExifData['GPSAltitudeRef'] )
 404                 ) {
 405                         // We know altitude data is a <num>/<denom> from the validation
 406                         // functions ran earlier. But multiplying such a string by -1
 407                         // doesn't work well, so convert.
 408                         list( $num, $denom ) = explode( '/', $this->mFilteredExifData['GPSAltitude'] );
 409                         $this->mFilteredExifData['GPSAltitude'] = $num / $denom;
 410
 411                         if ( $this->mFilteredExifData['GPSAltitudeRef'] === "\1" ) {
 412                                 $this->mFilteredExifData['GPSAltitude'] *= -1;
 413                         }
 414                         unset( $this->mFilteredExifData['GPSAltitudeRef'] );
 415                 }
 416
 417                 $this->exifPropToOrd( 'FileSource' );
 418                 $this->exifPropToOrd( 'SceneType' );
 419
 420                 $this->charCodeString( 'UserComment' );
 421                 $this->charCodeString( 'GPSProcessingMethod' );
 422                 $this->charCodeString( 'GPSAreaInformation' );
 423
 424                 //ComponentsConfiguration should really be an array instead of a string...
 425                 //This turns a string of binary numbers into an array of numbers.
 426
 427                 if ( isset( $this->mFilteredExifData['ComponentsConfiguration'] ) ) {
 428                         $val = $this->mFilteredExifData['ComponentsConfiguration'];
 429                         $ccVals = array();
 430
 431                         $strLen = strlen( $val );
 432                         for ( $i = 0; $i < $strLen; $i++ ) {
 433                                 $ccVals[$i] = ord( substr( $val, $i, 1 ) );
 434                         }
 435                         $ccVals['_type'] = 'ol'; //this is for formatting later.
 436                         $this->mFilteredExifData['ComponentsConfiguration'] = $ccVals;
 437                 }
 438
 439                 //GPSVersion(ID) is treated as the wrong type by php exif support.
 440                 //Go through each byte turning it into a version string.
 441                 //For example: "\x02\x02\x00\x00" -> "2.2.0.0"
 442
 443                 //Also change exif tag name from GPSVersion (what php exif thinks it is)
 444                 //to GPSVersionID (what the exif standard thinks it is).
 445
 446                 if ( isset( $this->mFilteredExifData['GPSVersion'] ) ) {
 447                         $val = $this->mFilteredExifData['GPSVersion'];
 448                         $newVal = '';
 449
 450                         $strLen = strlen( $val );
 451                         for ( $i = 0; $i < $strLen; $i++ ) {
 452                                 if ( $i !== 0 ) {
 453                                         $newVal .= '.';
 454                                 }
 455                                 $newVal .= ord( substr( $val, $i, 1 ) );
 456                         }
 457
 458                         if ( $this->byteOrder === 'LE' ) {
 459                                 // Need to reverse the string
 460                                 $newVal2 = '';
 461                                 for ( $i = strlen( $newVal ) - 1; $i >= 0; $i-- ) {
 462                                         $newVal2 .= substr( $newVal, $i, 1 );
 463                                 }
 464                                 $this->mFilteredExifData['GPSVersionID'] = $newVal2;
 465                         } else {
 466                                 $this->mFilteredExifData['GPSVersionID'] = $newVal;
 467                         }
 468                         unset( $this->mFilteredExifData['GPSVersion'] );
 469                 }
 470         }
 471
 472         /**
 473          * Do userComment tags and similar. See pg. 34 of exif standard.
 474          * basically first 8 bytes is charset, rest is value.
 475          * This has not been tested on any shift-JIS strings.
 476          * @param string $prop prop name.
 477          */
 478         private function charCodeString( $prop ) {
 479                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 480
 481                         if ( strlen( $this->mFilteredExifData[$prop] ) <= 8 ) {
 482                                 //invalid. Must be at least 9 bytes long.
 483
 484                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, false );
 485                                 unset( $this->mFilteredExifData[$prop] );
 486
 487                                 return;
 488                         }
 489                         $charCode = substr( $this->mFilteredExifData[$prop], 0, 8 );
 490                         $val = substr( $this->mFilteredExifData[$prop], 8 );
 491
 492                         switch ( $charCode ) {
 493                                 case "\x4A\x49\x53\x00\x00\x00\x00\x00":
 494                                         //JIS
 495                                         $charset = "Shift-JIS";
 496                                         break;
 497                                 case "UNICODE\x00":
 498                                         $charset = "UTF-16" . $this->byteOrder;
 499                                         break;
 500                                 default: //ascii or undefined.
 501                                         $charset = "";
 502                                         break;
 503                         }
 504                         // This could possibly check to see if iconv is really installed
 505                         // or if we're using the compatibility wrapper in globalFunctions.php
 506                         if ( $charset ) {
 507                                 wfSuppressWarnings();
 508                                 $val = iconv( $charset, 'UTF-8//IGNORE', $val );
 509                                 wfRestoreWarnings();
 510                         } else {
 511                                 // if valid utf-8, assume that, otherwise assume windows-1252
 512                                 $valCopy = $val;
 513                                 UtfNormal::quickIsNFCVerify( $valCopy ); //validates $valCopy.
 514                                 if ( $valCopy !== $val ) {
 515                                         wfSuppressWarnings();
 516                                         $val = iconv( 'Windows-1252', 'UTF-8//IGNORE', $val );
 517                                         wfRestoreWarnings();
 518                                 }
 519                         }
 520
 521                         //trim and check to make sure not only whitespace.
 522                         $val = trim( $val );
 523                         if ( strlen( $val ) === 0 ) {
 524                                 //only whitespace.
 525                                 $this->debug( $this->mFilteredExifData[$prop], __FUNCTION__, "$prop: Is only whitespace" );
 526                                 unset( $this->mFilteredExifData[$prop] );
 527
 528                                 return;
 529                         }
 530
 531                         //all's good.
 532                         $this->mFilteredExifData[$prop] = $val;
 533                 }
 534         }
 535
 536         /**
 537          * Convert an Exif::UNDEFINED from a raw binary string
 538          * to its value. This is sometimes needed depending on
 539          * the type of UNDEFINED field
 540          * @param string $prop name of property
 541          */
 542         private function exifPropToOrd( $prop ) {
 543                 if ( isset( $this->mFilteredExifData[$prop] ) ) {
 544                         $this->mFilteredExifData[$prop] = ord( $this->mFilteredExifData[$prop] );
 545                 }
 546         }
 547
 548         /**
 549          * Convert gps in exif form to a single floating point number
 550          * for example 10 degress 20`40`` S -> -10.34444
 551          * @param string $prop a gps coordinate exif tag name (like GPSLongitude)
 552          */
 553         private function exifGPStoNumber( $prop ) {
 554                 $loc =& $this->mFilteredExifData[$prop];
 555                 $dir =& $this->mFilteredExifData[$prop . 'Ref'];
 556                 $res = false;
 557
 558                 if ( isset( $loc ) && isset( $dir )
 559                         && ( $dir === 'N' || $dir === 'S' || $dir === 'E' || $dir === 'W' )
 560                 ) {
 561                         list( $num, $denom ) = explode( '/', $loc[0] );
 562                         $res = $num / $denom;
 563                         list( $num, $denom ) = explode( '/', $loc[1] );
 564                         $res += ( $num / $denom ) * ( 1 / 60 );
 565                         list( $num, $denom ) = explode( '/', $loc[2] );
 566                         $res += ( $num / $denom ) * ( 1 / 3600 );
 567
 568                         if ( $dir === 'S' || $dir === 'W' ) {
 569                                 $res *= -1; // make negative
 570                         }
 571                 }
 572
 573                 // update the exif records.
 574
 575                 if ( $res !== false ) { // using !== as $res could potentially be 0
 576                         $this->mFilteredExifData[$prop] = $res;
 577                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 578                 } else { // if invalid
 579                         unset( $this->mFilteredExifData[$prop] );
 580                         unset( $this->mFilteredExifData[$prop . 'Ref'] );
 581                 }
 582         }
 583
 584         /**
 585          * Use FormatMetadata to create formatted values for display to user
 586          * (is this ever used?)
 587          *
 588          * @deprecated since 1.18
 589          */
 590         function makeFormattedData() {
 591                 wfDeprecated( __METHOD__, '1.18' );
 592                 $this->mFormattedExifData = FormatMetadata::getFormattedData(
 593                         $this->mFilteredExifData );
 594         }
 595
 596         /**#@-*/
 597
 598         /**#@+
 599          * @return array
 600          */
 601         /**
 602          * Get $this->mRawExifData
 603          * @return array
 604          */
 605         function getData() {
 606                 return $this->mRawExifData;
 607         }
 608
 609         /**
 610          * Get $this->mFilteredExifData
 611          */
 612         function getFilteredData() {
 613                 return $this->mFilteredExifData;
 614         }
 615
 616         /**
 617          * Get $this->mFormattedExifData
 618          *
 619          * This returns the data for display to user.
 620          * Its unclear if this is ever used.
 621          *
 622          * @deprecated since 1.18
 623          */
 624         function getFormattedData() {
 625                 wfDeprecated( __METHOD__, '1.18' );
 626                 if ( !$this->mFormattedExifData ) {
 627                         $this->makeFormattedData();
 628                 }
 629
 630                 return $this->mFormattedExifData;
 631         }
 632
 633         /**#@-*/
 634
 635         /**
 636          * The version of the output format
 637          *
 638          * Before the actual metadata information is saved in the database we
 639          * strip some of it since we don't want to save things like thumbnails
 640          * which usually accompany Exif data. This value gets saved in the
 641          * database along with the actual Exif data, and if the version in the
 642          * database doesn't equal the value returned by this function the Exif
 643          * data is regenerated.
 644          *
 645          * @return int
 646          */
 647         public static function version() {
 648                 return 2; // We don't need no bloddy constants!
 649         }
 650
 651         /**#@+
 652          * Validates if a tag value is of the type it should be according to the Exif spec
 653          *
 654          * @private
 655          *
 656          * @param $in Mixed: the input value to check
 657          * @return bool
 658          */
 659         private function isByte( $in ) {
 660                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 255 ) {
 661                         $this->debug( $in, __FUNCTION__, true );
 662
 663                         return true;
 664                 } else {
 665                         $this->debug( $in, __FUNCTION__, false );
 666
 667                         return false;
 668                 }
 669         }
 670
 671         /**
 672          * @param $in
 673          * @return bool
 674          */
 675         private function isASCII( $in ) {
 676                 if ( is_array( $in ) ) {
 677                         return false;
 678                 }
 679
 680                 if ( preg_match( "/[^\x0a\x20-\x7e]/", $in ) ) {
 681                         $this->debug( $in, __FUNCTION__, 'found a character not in our whitelist' );
 682
 683                         return false;
 684                 }
 685
 686                 if ( preg_match( '/^\s*$/', $in ) ) {
 687                         $this->debug( $in, __FUNCTION__, 'input consisted solely of whitespace' );
 688
 689                         return false;
 690                 }
 691
 692                 return true;
 693         }
 694
 695         /**
 696          * @param $in
 697          * @return bool
 698          */
 699         private function isShort( $in ) {
 700                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 65536 ) {
 701                         $this->debug( $in, __FUNCTION__, true );
 702
 703                         return true;
 704                 } else {
 705                         $this->debug( $in, __FUNCTION__, false );
 706
 707                         return false;
 708                 }
 709         }
 710
 711         /**
 712          * @param $in
 713          * @return bool
 714          */
 715         private function isLong( $in ) {
 716                 if ( !is_array( $in ) && sprintf( '%d', $in ) == $in && $in >= 0 && $in <= 4294967296 ) {
 717                         $this->debug( $in, __FUNCTION__, true );
 718
 719                         return true;
 720                 } else {
 721                         $this->debug( $in, __FUNCTION__, false );
 722
 723                         return false;
 724                 }
 725         }
 726
 727         /**
 728          * @param $in
 729          * @return bool
 730          */
 731         private function isRational( $in ) {
 732                 $m = array();
 733
 734                 # Avoid division by zero
 735                 if ( !is_array( $in )
 736                         && preg_match( '/^(\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 737                 ) {
 738                         return $this->isLong( $m[1] ) && $this->isLong( $m[2] );
 739                 } else {
 740                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 741
 742                         return false;
 743                 }
 744         }
 745
 746         /**
 747          * @param $in
 748          * @return bool
 749          */
 750         private function isUndefined( $in ) {
 751                 $this->debug( $in, __FUNCTION__, true );
 752
 753                 return true;
 754         }
 755
 756         /**
 757          * @param $in
 758          * @return bool
 759          */
 760         private function isSlong( $in ) {
 761                 if ( $this->isLong( abs( $in ) ) ) {
 762                         $this->debug( $in, __FUNCTION__, true );
 763
 764                         return true;
 765                 } else {
 766                         $this->debug( $in, __FUNCTION__, false );
 767
 768                         return false;
 769                 }
 770         }
 771
 772         /**
 773          * @param $in
 774          * @return bool
 775          */
 776         private function isSrational( $in ) {
 777                 $m = array();
 778
 779                 # Avoid division by zero
 780                 if ( !is_array( $in ) &&
 781                         preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $in, $m )
 782                 ) {
 783                         return $this->isSlong( $m[0] ) && $this->isSlong( $m[1] );
 784                 } else {
 785                         $this->debug( $in, __FUNCTION__, 'fed a non-fraction value' );
 786
 787                         return false;
 788                 }
 789         }
 790
 791         /**#@-*/
 792
 793         /**
 794          * Validates if a tag has a legal value according to the Exif spec
 795          *
 796          * @private
 797          * @param string $section section where tag is located.
 798          * @param string $tag the tag to check.
 799          * @param $val Mixed: the value of the tag.
 800          * @param $recursive Boolean: true if called recursively for array types.
 801          * @return bool
 802          */
 803         private function validate( $section, $tag, $val, $recursive = false ) {
 804                 $debug = "tag is '$tag'";
 805                 $etype = $this->mExifTags[$section][$tag];
 806                 $ecount = 1;
 807                 if ( is_array( $etype ) ) {
 808                         list( $etype, $ecount ) = $etype;
 809                         if ( $recursive ) {
 810                                 $ecount = 1; // checking individual elements
 811                         }
 812                 }
 813                 $count = count( $val );
 814                 if ( $ecount != $count ) {
 815                         $this->debug( $val, __FUNCTION__, "Expected $ecount elements for $tag but got $count" );
 816
 817                         return false;
 818                 }
 819                 if ( $count > 1 ) {
 820                         foreach ( $val as $v ) {
 821                                 if ( !$this->validate( $section, $tag, $v, true ) ) {
 822                                         return false;
 823                                 }
 824                         }
 825
 826                         return true;
 827                 }
 828                 // Does not work if not typecast
 829                 switch ( (string)$etype ) {
 830                         case (string)Exif::BYTE:
 831                                 $this->debug( $val, __FUNCTION__, $debug );
 832
 833                                 return $this->isByte( $val );
 834                         case (string)Exif::ASCII:
 835                                 $this->debug( $val, __FUNCTION__, $debug );
 836
 837                                 return $this->isASCII( $val );
 838                         case (string)Exif::SHORT:
 839                                 $this->debug( $val, __FUNCTION__, $debug );
 840
 841                                 return $this->isShort( $val );
 842                         case (string)Exif::LONG:
 843                                 $this->debug( $val, __FUNCTION__, $debug );
 844
 845                                 return $this->isLong( $val );
 846                         case (string)Exif::RATIONAL:
 847                                 $this->debug( $val, __FUNCTION__, $debug );
 848
 849                                 return $this->isRational( $val );
 850                         case (string)Exif::SHORT_OR_LONG:
 851                                 $this->debug( $val, __FUNCTION__, $debug );
 852
 853                                 return $this->isShort( $val ) || $this->isLong( $val );
 854                         case (string)Exif::UNDEFINED:
 855                                 $this->debug( $val, __FUNCTION__, $debug );
 856
 857                                 return $this->isUndefined( $val );
 858                         case (string)Exif::SLONG:
 859                                 $this->debug( $val, __FUNCTION__, $debug );
 860
 861                                 return $this->isSlong( $val );
 862                         case (string)Exif::SRATIONAL:
 863                                 $this->debug( $val, __FUNCTION__, $debug );
 864
 865                                 return $this->isSrational( $val );
 866                         case (string)Exif::IGNORE:
 867                                 $this->debug( $val, __FUNCTION__, $debug );
 868
 869                                 return false;
 870                         default:
 871                                 $this->debug( $val, __FUNCTION__, "The tag '$tag' is unknown" );
 872
 873                                 return false;
 874                 }
 875         }
 876
 877         /**
 878          * Convenience function for debugging output
 879          *
 880          * @private
 881          *
 882          * @param $in Mixed:
 883          * @param $fname String:
 884          * @param $action Mixed: , default NULL.
 885          */
 886         private function debug( $in, $fname, $action = null ) {
 887                 if ( !$this->log ) {
 888                         return;
 889                 }
 890                 $type = gettype( $in );
 891                 $class = ucfirst( __CLASS__ );
 892                 if ( is_array( $in ) ) {
 893                         $in = print_r( $in, true );
 894                 }
 895
 896                 if ( $action === true ) {
 897                         wfDebugLog( $this->log, "$class::$fname: accepted: '$in' (type: $type)\n" );
 898                 } elseif ( $action === false ) {
 899                         wfDebugLog( $this->log, "$class::$fname: rejected: '$in' (type: $type)\n" );
 900                 } elseif ( $action === null ) {
 901                         wfDebugLog( $this->log, "$class::$fname: input was: '$in' (type: $type)\n" );
 902                 } else {
 903                         wfDebugLog( $this->log, "$class::$fname: $action (type: $type; content: '$in')\n" );
 904                 }
 905         }
 906
 907         /**
 908          * Convenience function for debugging output
 909          *
 910          * @private
 911          *
 912          * @param string $fname the name of the function calling this function
 913          * @param $io Boolean: Specify whether we're beginning or ending
 914          */
 915         private function debugFile( $fname, $io ) {
 916                 if ( !$this->log ) {
 917                         return;
 918                 }
 919                 $class = ucfirst( __CLASS__ );
 920                 if ( $io ) {
 921                         wfDebugLog( $this->log, "$class::$fname: begin processing: '{$this->basename}'\n" );
 922                 } else {
 923                         wfDebugLog( $this->log, "$class::$fname: end processing: '{$this->basename}'\n" );
 924                 }
 925         }
 926 }