includes/media/XMPValidate.php

   1 <?php
   2 /**
   3 * This contains some static methods for
   4 * validating XMP properties. See XMPInfo and XMPReader classes.
   5 *
   6 * Each of these functions take the same parameters
   7 * * an info array which is a subset of the XMPInfo::items array
   8 * * A value (passed as reference) to validate. This can be either a
   9 *       simple value or an array
  10 * * A boolean to determine if this is validating a simple or complex values
  11 *
  12 * It should be noted that when an array is being validated, typically the validation
  13 * function is called once for each value, and then once at the end for the entire array.
  14 *
  15 * These validation functions can also be used to modify the data. See the gps and flash one's
  16 * for example.
  17 *
  18 * @see http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart1.pdf starting at pg 28
  19 * @see http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart2.pdf starting at pg 11
  20 */
  21 class XMPValidate {
  22         /**
  23         * function to validate boolean properties ( True or False )
  24         *
  25         * @param $info Array information about current property
  26         * @param &$val Mixed current value to validate
  27         * @param $standalone Boolean if this is a simple property or array
  28         */
  29         public static function validateBoolean( $info, &$val, $standalone ) {
  30                 if ( !$standalone ) {
  31                         // this only validates standalone properties, not arrays, etc
  32                         return;
  33                 }
  34                 if ( $val !== 'True' && $val !== 'False' ) {
  35                         wfDebugLog( 'XMP', __METHOD__ . " Expected True or False but got $val" );
  36                         $val = null;
  37                 }
  38
  39         }
  40         /**
  41         * function to validate rational properties ( 12/10 )
  42         *
  43         * @param $info Array information about current property
  44         * @param &$val Mixed current value to validate
  45         * @param $standalone Boolean if this is a simple property or array
  46         */
  47         public static function validateRational( $info, &$val, $standalone ) {
  48                 if ( !$standalone ) {
  49                         // this only validates standalone properties, not arrays, etc
  50                         return;
  51                 }
  52                 if ( !preg_match( '/^(?:-?\d+)\/(?:\d+[1-9]|[1-9]\d*)$/D', $val ) ) {
  53                         wfDebugLog( 'XMP', __METHOD__ . " Expected rational but got $val" );
  54                         $val = null;
  55                 }
  56
  57         }
  58         /**
  59         * function to validate rating properties -1, 0-5
  60         *
  61         * if its outside of range put it into range.
  62         *
  63         * @see MWG spec
  64         * @param $info Array information about current property
  65         * @param &$val Mixed current value to validate
  66         * @param $standalone Boolean if this is a simple property or array
  67         */
  68         public static function validateRating( $info, &$val, $standalone ) {
  69                 if ( !$standalone ) {
  70                         // this only validates standalone properties, not arrays, etc
  71                         return;
  72                 }
  73                 if ( !preg_match( '/^[-+]?\d*(?:\.?\d*)$/D', $val )
  74                         || !is_numeric($val)
  75                  ) {
  76                         wfDebugLog( 'XMP', __METHOD__ . " Expected rating but got $val" );
  77                         $val = null;
  78                         return;
  79                 } else {
  80                         $nVal = (float) $val;
  81                         if ( $nVal < 0 ) {
  82                                 // We do < 0 here instead of < -1 here, since
  83                                 // the values between 0 and -1 are also illegal
  84                                 // as -1 is meant as a special reject rating.
  85                                 wfDebugLog( 'XMP', __METHOD__ . " Rating too low, setting to -1 (Rejected)");
  86                                 $val = '-1';
  87                                 return;
  88                         }
  89                         if ( $nVal > 5 ) {
  90                                 wfDebugLog( 'XMP', __METHOD__ . " Rating too high, setting to 5");
  91                                 $val = '5';
  92                                 return;
  93                         }
  94                 }
  95         }
  96         /**
  97         * function to validate integers
  98         *
  99         * @param $info Array information about current property
 100         * @param &$val Mixed current value to validate
 101         * @param $standalone Boolean if this is a simple property or array
 102         */
 103         public static function validateInteger( $info, &$val, $standalone ) {
 104                 if ( !$standalone ) {
 105                         // this only validates standalone properties, not arrays, etc
 106                         return;
 107                 }
 108                 if ( !preg_match( '/^[-+]?\d+$/D', $val ) ) {
 109                         wfDebugLog( 'XMP', __METHOD__ . " Expected integer but got $val" );
 110                         $val = null;
 111                 }
 112
 113         }
 114         /**
 115         * function to validate properties with a fixed number of allowed
 116         * choices. (closed choice)
 117         *
 118         * @param $info Array information about current property
 119         * @param &$val Mixed current value to validate
 120         * @param $standalone Boolean if this is a simple property or array
 121         */
 122         public static function validateClosed( $info, &$val, $standalone ) {
 123                 if ( !$standalone ) {
 124                         // this only validates standalone properties, not arrays, etc
 125                         return;
 126                 }
 127
 128                 //check if its in a numeric range
 129                 $inRange = false;
 130                 if ( isset( $info['rangeLow'] )
 131                         && isset( $info['rangeHigh'] )
 132                         && is_numeric( $val )
 133                         && ( intval( $val ) <= $info['rangeHigh'] )
 134                         && ( intval( $val ) >= $info['rangeLow'] )
 135                 ) {
 136                         $inRange = true;
 137                 }
 138
 139                 if ( !isset( $info['choices'][$val] ) && !$inRange ) {
 140                         wfDebugLog( 'XMP', __METHOD__ . " Expected closed choice, but got $val" );
 141                         $val = null;
 142                 }
 143         }
 144         /**
 145         * function to validate and modify flash structure
 146         *
 147         * @param $info Array information about current property
 148         * @param &$val Mixed current value to validate
 149         * @param $standalone Boolean if this is a simple property or array
 150         */
 151         public static function validateFlash( $info, &$val, $standalone ) {
 152                 if ( $standalone ) {
 153                         // this only validates flash structs, not individual properties
 154                         return;
 155                 }
 156                 if ( !( isset( $val['Fired'] )
 157                         && isset( $val['Function'] )
 158                         && isset( $val['Mode'] )
 159                         && isset( $val['RedEyeMode'] )
 160                         && isset( $val['Return'] )
 161                 ) ) {
 162                         wfDebugLog( 'XMP', __METHOD__ . " Flash structure did not have all the required components" );
 163                         $val = null;
 164                 } else {
 165                         $val = ( "\0" | ( $val['Fired'] === 'True' )
 166                                 | ( intval( $val['Return'] ) << 1 )
 167                                 | ( intval( $val['Mode'] ) << 3 )
 168                                 | ( ( $val['Function'] === 'True' ) << 5 )
 169                                 | ( ( $val['RedEyeMode'] === 'True' ) << 6 ) );
 170                 }
 171         }
 172         /**
 173         * function to validate LangCode properties ( en-GB, etc )
 174         *
 175         * This is just a naive check to make sure it somewhat looks like a lang code.
 176         *
 177         * @see rfc 3066
 178         * @see http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart1.pdf page 30 (section 8.2.2.5)
 179         *
 180         * @param $info Array information about current property
 181         * @param &$val Mixed current value to validate
 182         * @param $standalone Boolean if this is a simple property or array
 183         */
 184         public static function validateLangCode( $info, &$val, $standalone ) {
 185                 if ( !$standalone ) {
 186                         // this only validates standalone properties, not arrays, etc
 187                         return;
 188                 }
 189                 if ( !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $val) ) {
 190                         //this is a rather naive check.
 191                         wfDebugLog( 'XMP', __METHOD__ . " Expected Lang code but got $val" );
 192                         $val = null;
 193                 }
 194
 195         }
 196         /**
 197         * function to validate date properties, and convert to Exif format.
 198         *
 199         * @param $info Array information about current property
 200         * @param &$val Mixed current value to validate. Converts to TS_EXIF as a side-effect.
 201         * @param $standalone Boolean if this is a simple property or array
 202         */
 203         public static function validateDate( $info, &$val, $standalone ) {
 204                 if ( !$standalone ) {
 205                         // this only validates standalone properties, not arrays, etc
 206                         return;
 207                 }
 208                 $res = array();
 209                 if ( !preg_match(
 210                         /* ahh! scary regex... */
 211                         '/^([0-3]\d{3})(?:-([01]\d)(?:-([0-3]\d)(?:T([0-2]\d):([0-6]\d)(?::([0-6]\d)(?:\.\d+)?)?([-+]\d{2}:\d{2}|Z)?)?)?)?$/D'
 212                         , $val, $res)
 213                 ) {
 214                         wfDebugLog( 'XMP', __METHOD__ . " Expected date but got $val" );
 215                         $val = null;
 216                 } else {
 217                         /*
 218                          * $res is formatted as follows:
 219                          * 0 -> full date.
 220                          * 1 -> year, 2-> month, 3-> day, 4-> hour, 5-> minute, 6->second
 221                          * 7-> Timezone specifier (Z or something like +12:30 )
 222                          * many parts are optional, some aren't. For example if you specify
 223                          * minute, you must specify hour, day, month, and year but not second or TZ.
 224                          */
 225
 226                         /*
 227                          * First of all, if year = 0000, Something is wrongish,
 228                          * so don't extract. This seems to happen when
 229                          * some programs convert between metadata formats.
 230                          */
 231                         if ( $res[1] === '0000' ) {
 232                                 wfDebugLog( 'XMP', __METHOD__ . " Invalid date (year 0): $val" );
 233                                 $val = null;
 234                                 return;
 235                         }
 236                         //if month, etc unspecified, full out as 01.
 237                         $res[2] = isset( $res[2] ) ? $res[2] : '01'; //month
 238                         $res[3] = isset( $res[3] ) ? $res[3] : '01'; //day
 239                         if ( !isset( $res[4] ) ) { //hour
 240                                 //just have the year month day
 241                                 $val = $res[1] . ':' . $res[2] . ':' . $res[3];
 242                                 return;
 243                         }
 244                         //if hour is set, so is minute or regex above will fail.
 245                         //Extra check for empty string necessary due to TZ but no second case.
 246                         $res[6] = isset( $res[6] ) && $res[6] != '' ? $res[6] : '00';
 247
 248                         if ( !isset( $res[7] ) || $res[7] === 'Z' ) {
 249                                 $val = $res[1] . ':' . $res[2] . ':' . $res[3]
 250                                         . ' ' . $res[4] . ':' . $res[5] . ':' . $res[6];
 251                                 return;
 252                         }
 253
 254                         //do timezone processing. We've already done the case that tz = Z.
 255
 256                         $unix = wfTimestamp( TS_UNIX, $res[1] . $res[2] . $res[3] . $res[4] . $res[5] . $res[6] );
 257                         $offset = intval( substr( $res[7], 1, 2 ) ) * 60 * 60;
 258                         $offset += intval( substr( $res[7], 4, 2 ) ) * 60;
 259                         if ( substr( $res[7], 0, 1 ) === '-' ) {
 260                                 $offset = -$offset;
 261                         }
 262                         $val = wfTimestamp( TS_EXIF, $unix + $offset );
 263                 }
 264
 265         }
 266         /** function to validate, and more importantly
 267          * translate the XMP DMS form of gps coords to
 268          * the decimal form we use.
 269          *
 270          * @see http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart2.pdf
 271          *        section 1.2.7.4 on page 23
 272          *
 273          * @param $info Array unused (info about prop)
 274          * @param &$val String GPS string in either DDD,MM,SSk or
 275          *           or DDD,MM.mmk form
 276          * @param $standalone Boolean if its a simple prop (should always be true)
 277          */
 278         public static function validateGPS ( $info, &$val, $standalone ) {
 279                 if ( !$standalone ) {
 280                         return;
 281                 }
 282
 283                 $m = array();
 284                 if ( preg_match(
 285                         '/(\d{1,3}),(\d{1,2}),(\d{1,2})([NWSE])/D',
 286                         $val, $m )
 287                 ) {
 288                         $coord = intval( $m[1] );
 289                         $coord += intval( $m[2] ) * (1/60);
 290                         $coord += intval( $m[3] ) * (1/3600);
 291                         if ( $m[4] === 'S' || $m[4] === 'W' ) {
 292                                 $coord = -$coord;
 293                         }
 294                         $val = $coord;
 295                         return;
 296                 } elseif ( preg_match(
 297                         '/(\d{1,3}),(\d{1,2}(?:.\d*)?)([NWSE])/D',
 298                         $val, $m )
 299                 ) {
 300                         $coord = intval( $m[1] );
 301                         $coord += floatval( $m[2] ) * (1/60);
 302                         if ( $m[3] === 'S' || $m[3] === 'W' ) {
 303                                 $coord = -$coord;
 304                         }
 305                         $val = $coord;
 306                         return;
 307
 308                 } else {
 309                         wfDebugLog( 'XMP', __METHOD__
 310                                 . " Expected GPSCoordinate, but got $val." );
 311                         $val = null;
 312                         return;
 313                 }
 314
 315         }
 316
 317 }