includes/media/XMP.php

   1 <?php
   2 /**
   3 * Class for reading xmp data containing properties relevant to
   4 * images, and spitting out an array that FormatExif accepts.
   5 *
   6 * Note, this is not meant to recognize every possible thing you can
   7 * encode in XMP. It should recognize all the properties we want.
   8 * For example it doesn't have support for structures with multiple
   9 * nesting levels, as none of the properties we're supporting use that
  10 * feature. If it comes across properties it doesn't recognize, it should
  11 * ignore them.
  12 *
  13 * The public methods one would call in this class are
  14 * - parse( $content )
  15 *       Reads in xmp content.
  16 *       Can potentially be called multiple times with partial data each time.
  17 * - parseExtended( $content )
  18 *       Reads XMPExtended blocks (jpeg files only).
  19 * - getResults
  20 *       Outputs a results array.
  21 *
  22 * Note XMP kind of looks like rdf. They are not the same thing - XMP is
  23 * encoded as a specific subset of rdf. This class can read XMP. It cannot
  24 * read rdf.
  25 *
  26 */
  27 class XMPReader {
  28
  29         private $curItem = array();        // array to hold the current element (and previous element, and so on)
  30         private $ancestorStruct = false;   // the structure name when processing nested structures.
  31         private $charContent = false;      // temporary holder for character data that appears in xmp doc.
  32         private $mode = array();           // stores the state the xmpreader is in (see MODE_FOO constants)
  33         private $results = array();        // array to hold results
  34         private $processingArray = false;  // if we're doing a seq or bag.
  35         private $itemLang = false;         // used for lang alts only
  36
  37         private $xmlParser;
  38         private $charset = false;
  39         private $extendedXMPOffset = 0;
  40
  41         protected $items;
  42
  43         /*
  44         * These are various mode constants.
  45         * they are used to figure out what to do
  46         * with an element when its encountered.
  47         *
  48         * For example, MODE_IGNORE is used when processing
  49         * a property we're not interested in. So if a new
  50         * element pops up when we're in that mode, we ignore it.
  51         */
  52         const MODE_INITIAL = 0;
  53         const MODE_IGNORE  = 1;
  54         const MODE_LI      = 2;
  55         const MODE_LI_LANG = 3;
  56         const MODE_QDESC   = 4;
  57
  58         // The following MODE constants are also used in the
  59         // $items array to denote what type of property the item is.
  60         const MODE_SIMPLE    = 10;
  61         const MODE_STRUCT    = 11; // structure (associative array)
  62         const MODE_SEQ       = 12; // ordered list
  63         const MODE_BAG       = 13; // unordered list
  64         const MODE_LANG      = 14;
  65         const MODE_ALT       = 15; // non-language alt. Currently not implemented, and not needed atm.
  66         const MODE_BAGSTRUCT = 16; // A BAG of Structs.
  67
  68         const NS_RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
  69         const NS_XML = 'http://www.w3.org/XML/1998/namespace';
  70
  71
  72         /**
  73         * Constructor.
  74         *
  75         * Primary job is to initialize the XMLParser
  76         */
  77         function __construct() {
  78
  79                 if ( !function_exists( 'xml_parser_create_ns' ) ) {
  80                         // this should already be checked by this point
  81                         throw new MWException( 'XMP support requires XML Parser' );
  82                 }
  83
  84                 $this->items = XMPInfo::getItems();
  85
  86                 $this->resetXMLParser();
  87
  88         }
  89         /**
  90         * Main use is if a single item has multiple xmp documents describing it.
  91         * For example in jpeg's with extendedXMP
  92         */
  93         private function resetXMLParser() {
  94
  95                 if ($this->xmlParser) {
  96                         //is this needed?
  97                         xml_parser_free( $this->xmlParser );
  98                 }
  99
 100                 $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' );
 101                 xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 );
 102                 xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 );
 103
 104                 xml_set_element_handler( $this->xmlParser,
 105                         array( $this, 'startElement' ),
 106                         array( $this, 'endElement' ) );
 107
 108                 xml_set_character_data_handler( $this->xmlParser, array( $this, 'char' ) );
 109
 110
 111         }
 112
 113         /** Destroy the xml parser
 114         *
 115         * Not sure if this is actually needed.
 116         */
 117         function __destruct() {
 118                 // not sure if this is needed.
 119                 xml_parser_free( $this->xmlParser );
 120         }
 121
 122         /** Get the result array. Do some post-processing before returning
 123         * the array, and transform any metadata that is special-cased.
 124         *
 125         * @return Array array of results as an array of arrays suitable for
 126         *       FormatMetadata::getFormattedData().
 127         */
 128         public function getResults() {
 129                 // xmp-special is for metadata that affects how stuff
 130                 // is extracted. For example xmpNote:HasExtendedXMP.
 131
 132                 // It is also used to handle photoshop:AuthorsPosition
 133                 // which is weird and really part of another property,
 134                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 135                 // The location fields also use it.
 136
 137                 $data = $this->results;
 138
 139                 wfRunHooks('XMPGetResults', Array(&$data));
 140
 141                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 142                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 143                         && isset( $data['xmp-general']['Artist'][0] )
 144                 ) {
 145                         // Note, if there is more than one creator,
 146                         // this only applies to first. This also will
 147                         // only apply to the dc:Creator prop, not the
 148                         // exif:Artist prop.
 149
 150                         $data['xmp-general']['Artist'][0] =
 151                                 $data['xmp-special']['AuthorsPosition'] . ', '
 152                                 . $data['xmp-general']['Artist'][0];
 153                 }
 154
 155                 // Go through the LocationShown and LocationCreated
 156                 // changing it to the non-hierarchal form used by
 157                 // the other location fields.
 158
 159                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 160                         && is_array( $data['xmp-special']['LocationShown'][0] )
 161                 ) {
 162                         // the is_array is just paranoia. It should always
 163                         // be an array.
 164                         foreach( $data['xmp-special']['LocationShown'] as $loc ) {
 165                                 if ( !is_array( $loc ) ) {
 166                                         // To avoid copying over the _type meta-fields.
 167                                         continue;
 168                                 }
 169                                 foreach( $loc as $field => $val ) {
 170                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 171                                 }
 172                         }
 173                 }
 174                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 175                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 176                 ) {
 177                         // the is_array is just paranoia. It should always
 178                         // be an array.
 179                         foreach( $data['xmp-special']['LocationCreated'] as $loc ) {
 180                                 if ( !is_array( $loc ) ) {
 181                                         // To avoid copying over the _type meta-fields.
 182                                         continue;
 183                                 }
 184                                 foreach(  $loc as $field => $val ) {
 185                                         $data['xmp-general'][$field . 'Created'][] = $val;
 186                                 }
 187                         }
 188                 }
 189
 190
 191                 // We don't want to return the special values, since they're
 192                 // special and not info to be stored about the file.
 193                 unset( $data['xmp-special'] );
 194
 195                 // Convert GPSAltitude to negative if below sea level.
 196                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] ) ) {
 197                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1'
 198                                 && isset( $data['xmp-exif']['GPSAltitude'] )
 199                         ) {
 200                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 201                         }
 202                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 203                 }
 204
 205                 return $data;
 206         }
 207
 208         /**
 209         * Main function to call to parse XMP. Use getResults to
 210         * get results.
 211         *
 212         * Also catches any errors during processing, writes them to
 213         * debug log, blanks result array and returns false.
 214         *
 215         * @param String: $content XMP data
 216         * @param Boolean: $allOfIt If this is all the data (true) or if its split up (false). Default true
 217         * @param Boolean: $reset - does xml parser need to be reset. Default false
 218         * @return Boolean success.
 219         */
 220         public function parse( $content, $allOfIt = true, $reset = false ) {
 221                 if ( $reset ) {
 222                         $this->resetXMLParser();
 223                 }
 224                 try {
 225
 226                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 227                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 228                         if ( !$this->charset ) {
 229                                 $bom = array();
 230                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 231                                          $content, $bom )
 232                                 ) {
 233                                         switch ( $bom[0] ) {
 234                                                 case "\xFE\xFF":
 235                                                         $this->charset = 'UTF-16BE';
 236                                                         break;
 237                                                 case "\xFF\xFE":
 238                                                         $this->charset = 'UTF-16LE';
 239                                                         break;
 240                                                 case "\x00\x00\xFE\xFF":
 241                                                         $this->charset = 'UTF-32BE';
 242                                                         break;
 243                                                 case "\xFF\xFE\x00\x00":
 244                                                         $this->charset = 'UTF-32LE';
 245                                                         break;
 246                                                 case "\xEF\xBB\xBF":
 247                                                         $this->charset = 'UTF-8';
 248                                                         break;
 249                                                 default:
 250                                                         //this should be impossible to get to
 251                                                         throw new MWException("Invalid BOM");
 252                                                         break;
 253
 254                                         }
 255
 256                                 } else {
 257                                         // standard specifically says, if no bom assume utf-8
 258                                         $this->charset = 'UTF-8';
 259                                 }
 260                         }
 261                         if ( $this->charset !== 'UTF-8' ) {
 262                                 //don't convert if already utf-8
 263                                 wfSuppressWarnings();
 264                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 265                                 wfRestoreWarnings();
 266                         }
 267
 268                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 269                         if ( !$ok ) {
 270                                 $error = xml_error_string( xml_get_error_code( $this->xmlParser ) );
 271                                 $where = 'line: ' . xml_get_current_line_number( $this->xmlParser )
 272                                         . ' column: ' . xml_get_current_column_number( $this->xmlParser )
 273                                         . ' byte offset: ' . xml_get_current_byte_index( $this->xmlParser );
 274
 275                                 wfDebugLog( 'XMP', "XMPReader::parse : Error reading XMP content: $error ($where)" );
 276                                 $this->results = array(); // blank if error.
 277                                 return false;
 278                         }
 279                 } catch ( MWException $e ) {
 280                         wfDebugLog( 'XMP', 'XMP parse error: ' . $e );
 281                         $this->results = array();
 282                         return false;
 283                 }
 284                 return true;
 285         }
 286
 287         /** Entry point for XMPExtended blocks in jpeg files
 288          *
 289          * @todo In serious need of testing
 290          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 291          * @param String $content XMPExtended block minus the namespace signature
 292          * @return Boolean If it succeeded.
 293          */
 294         public function parseExtended( $content ) {
 295                 // FIXME: This is untested. Hard to find example files
 296                 // or programs that make such files..
 297                 $guid = substr( $content, 0, 32 );
 298                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 299                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid )
 300                 {
 301                         wfDebugLog('XMP', __METHOD__ . " Ignoring XMPExtended block due to wrong guid (guid= '$guid' )");
 302                         return;
 303                 }
 304                 $len  = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 305
 306                 if (!$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 307                         wfDebugLog('XMP', __METHOD__ . 'Error reading extended XMP block, invalid length or offset.');
 308                         return false;
 309                 }
 310
 311
 312                 // we're not very robust here. we should accept it in the wrong order. To quote
 313                 // the xmp standard:
 314                 // "A JPEG writer should write the ExtendedXMP marker segments in order, immediately following the
 315                 // StandardXMP. However, the JPEG standard does not require preservation of marker segment order. A
 316                 // robust JPEG reader should tolerate the marker segments in any order."
 317                 //
 318                 // otoh the probability that an image will have more than 128k of metadata is rather low...
 319                 // so the probability that it will have > 128k, and be in the wrong order is very low...
 320
 321                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 322                         wfDebugLog('XMP', __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 323                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')');
 324                         return false;
 325                 }
 326
 327                 if ( $len['offset'] === 0 ) {
 328                         // if we're starting the extended block, we've probably already
 329                         // done the XMPStandard block, so reset.
 330                         $this->resetXMLParser();
 331                 }
 332
 333                 $this->extendedXMPOffset += $len['length'];
 334
 335                 $actualContent = substr( $content, 40 );
 336
 337                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 338                         $atEnd = true;
 339                 } else {
 340                         $atEnd = false;
 341                 }
 342
 343                 wfDebugLog('XMP', __METHOD__ . 'Parsing a XMPExtended block');
 344                 return $this->parse( $actualContent, $atEnd );
 345         }
 346
 347         /**
 348         * Character data handler
 349         * Called whenever character data is found in the xmp document.
 350         *
 351         * does nothing if we're in MODE_IGNORE or if the data is whitespace
 352         * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 353         * data in the other modes).
 354         *
 355         * As an example, this happens when we encounter XMP like:
 356         * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 357         * and are processing the 0/10 bit.
 358         *
 359         * @param $parser XMLParser reference to the xml parser
 360         * @param $data String Character data
 361         * @throws MWException on invalid data
 362         */
 363         function char( $parser, $data ) {
 364
 365                 $data = trim( $data );
 366                 if ( trim( $data ) === "" ) {
 367                         return;
 368                 }
 369
 370                 if ( !isset( $this->mode[0] ) ) {
 371                         throw new MWException( 'Unexpected character data before first rdf:Description element' );
 372                 }
 373
 374                 if ( $this->mode[0] === self::MODE_IGNORE ) return;
 375
 376                 if ( $this->mode[0] !== self::MODE_SIMPLE
 377                         && $this->mode[0] !== self::MODE_QDESC
 378                 ) {
 379                         throw new MWException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 380                 }
 381
 382                 // to check, how does this handle w.s.
 383                 if ( $this->charContent === false ) {
 384                         $this->charContent = $data;
 385                 } else {
 386                         $this->charContent .= $data;
 387                 }
 388
 389         }
 390         /** When we hit a closing element in MODE_IGNORE
 391         * Check to see if this is the element we started to ignore,
 392         * in which case we get out of MODE_IGNORE
 393         *
 394         * @param $elm String Namespace of element followed by a space and then tag name of element.
 395         */
 396         private function endElementModeIgnore ( $elm ) {
 397
 398                 if ( $this->curItem[0] === $elm ) {
 399                         array_shift( $this->curItem );
 400                         array_shift( $this->mode );
 401                 }
 402                 return;
 403
 404         }
 405         /**
 406         * Hit a closing element when in MODE_SIMPLE.
 407         * This generally means that we finished processing a
 408         * property value, and now have to save the result to the
 409         * results array
 410         *
 411         * For example, when processing:
 412         * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 413         * this deals with when we hit </exif:DigitalZoomRatio>.
 414         *
 415         * Or it could be if we hit the end element of a property
 416         * of a compound data structure (like a member of an array).
 417         *
 418         * @param $elm String namespace, space, and tag name.
 419         */
 420         private function endElementModeSimple ( $elm ) {
 421                 if ( $this->charContent !== false ) {
 422                         if ( $this->processingArray ) {
 423                                 // if we're processing an array, use the original element
 424                                 // name instead of rdf:li.
 425                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 426                         } else {
 427                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 428                         }
 429                         $this->saveValue( $ns, $tag, $this->charContent );
 430
 431                         $this->charContent = false; // reset
 432                 }
 433                 array_shift( $this->curItem );
 434                 array_shift( $this->mode );
 435
 436         }
 437         /**
 438         * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 439         * generally means we've finished processing a nested structure.
 440         * resets some internal variables to indicate that.
 441         *
 442         * Note this means we hit the </closing element> not the </rdf:Seq>.
 443         *
 444         * For example, when processing:
 445         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 446         *   </rdf:Seq> </exif:ISOSpeedRatings>
 447         *
 448         * This method is called when we hit the </exif:ISOSpeedRatings> tag.
 449         *
 450         * @param $elm String namespace . space . tag name.
 451         */
 452         private function endElementNested( $elm ) {
 453
 454                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 455                    in which case it could also be rdf:Description */
 456                 if ( $this->curItem[0] !== $elm
 457                         && !( $elm === self::NS_RDF . ' Description'
 458                                 && $this->mode[0] === self::MODE_STRUCT )
 459                  ) {
 460                         throw new MWException( "nesting mismatch. got a </$elm> but expected a </" . $this->curItem[0] . '>' );
 461                 }
 462
 463                 // Validate structures.
 464                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 465                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 466
 467                         $info =& $this->items[$ns][$tag];
 468                         $finalName = isset( $info['map_name'] )
 469                                 ? $info['map_name'] : $tag;
 470
 471                         $validate = is_array( $info['validate'] ) ? $info['validate']
 472                                 : array( 'XMPValidate', $info['validate'] );
 473
 474                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 475                                 // This can happen if all the members of the struct failed validation.
 476                                 wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> has no valid members." );
 477
 478                         } elseif ( is_callable( $validate ) ) {
 479                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 480                                 call_user_func_array( $validate, array( $info, &$val, false ) );
 481                                 if ( is_null( $val ) ) {
 482                                         // the idea being the validation function will unset the variable if
 483                                         // its invalid.
 484                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
 485                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 486                                 }
 487                         } else {
 488                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
 489                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 490                         }
 491                 }
 492
 493                 array_shift( $this->curItem );
 494                 array_shift( $this->mode );
 495                 $this->ancestorStruct = false;
 496                 $this->processingArray = false;
 497                 $this->itemLang = false;
 498         }
 499
 500         /**
 501         * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 502         * Add information about what type of element this is.
 503         *
 504         * Note we still have to hit the outer </property>
 505         *
 506         * For example, when processing:
 507         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 508         *   </rdf:Seq> </exif:ISOSpeedRatings>
 509         *
 510         * This method is called when we hit the </rdf:Seq>.
 511         * (For comparison, we call endElementModeSimple when we
 512         * hit the </rdf:li>)
 513         *
 514         * @param $elm String namespace . ' ' . element name
 515         */
 516         private function endElementModeLi( $elm ) {
 517
 518                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 519                 $info = $this->items[$ns][$tag];
 520                 $finalName = isset( $info['map_name'] )
 521                         ? $info['map_name'] : $tag;
 522
 523                 array_shift( $this->mode );
 524
 525                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 526                         wfDebugLog( 'XMP', __METHOD__ . " Empty compund element $finalName." );
 527                         return;
 528                 }
 529
 530                 if ( $elm === self::NS_RDF . ' Seq' ) {
 531                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 532                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 533                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 534                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 535                         // extra if needed as you could theoretically have a non-language alt.
 536                         if ( $info['mode'] === self::MODE_LANG ) {
 537                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 538                         }
 539
 540                 } else {
 541                         throw new MWException( __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
 542                 }
 543         }
 544         /**
 545         * End element while in MODE_QDESC
 546         * mostly when ending an element when we have a simple value
 547         * that has qualifiers.
 548         *
 549         * Qualifiers aren't all that common, and we don't do anything
 550         * with them.
 551         *
 552         * @param $elm String namespace and element
 553         */
 554         private function endElementModeQDesc( $elm ) {
 555
 556                 if ( $elm === self::NS_RDF . ' value' ) {
 557                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 558                         $this->saveValue( $ns, $tag, $this->charContent );
 559                         return;
 560                 } else {
 561                         array_shift( $this->mode );
 562                         array_shift( $this->curItem );
 563                 }
 564
 565
 566         }
 567         /**
 568         * Handler for hitting a closing element.
 569         *
 570         * generally just calls a helper function depending on what
 571         * mode we're in.
 572         *
 573         * Ignores the outer wrapping elements that are optional in
 574         * xmp and have no meaning.
 575         *
 576         * @param $parser XMLParser
 577         * @param $elm String namespace . ' ' . element name
 578         */
 579         function endElement( $parser, $elm ) {
 580                 if ( $elm === ( self::NS_RDF . ' RDF' )
 581                         || $elm === 'adobe:ns:meta/ xmpmeta'
 582                         || $elm === 'adobe:ns:meta/ xapmeta' )
 583                 {
 584                         // ignore these.
 585                         return;
 586                 }
 587
 588                 if ( $elm === self::NS_RDF . ' type' ) {
 589                         // these aren't really supported properly yet.
 590                         // However, it appears they almost never used.
 591                         wfDebugLog( 'XMP', __METHOD__ . ' encountered <rdf:type>' );
 592                 }
 593
 594                 if ( strpos( $elm, ' ' ) === false ) {
 595                         // This probably shouldn't happen.
 596                         // However, there is a bug in an adobe product
 597                         // that forgets the namespace on some things.
 598                         // (Luckily they are unimportant things).
 599                         wfDebugLog( 'XMP', __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 600                         return;
 601                 }
 602
 603                 if ( count( $this->mode[0] ) === 0 ) {
 604                         // This should never ever happen and means
 605                         // there is a pretty major bug in this class.
 606                         throw new MWException( 'Encountered end element with no mode' );
 607                 }
 608
 609                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 610                         // just to be paranoid. Should always have a curItem, except for initially
 611                         // (aka during MODE_INITAL).
 612                         throw new MWException( "Hit end element </$elm> but no curItem" );
 613                 }
 614
 615                 switch( $this->mode[0] ) {
 616                         case self::MODE_IGNORE:
 617                                 $this->endElementModeIgnore( $elm );
 618                                 break;
 619                         case self::MODE_SIMPLE:
 620                                 $this->endElementModeSimple( $elm );
 621                                 break;
 622                         case self::MODE_STRUCT:
 623                         case self::MODE_SEQ:
 624                         case self::MODE_BAG:
 625                         case self::MODE_LANG:
 626                         case self::MODE_BAGSTRUCT:
 627                                 $this->endElementNested( $elm );
 628                                 break;
 629                         case self::MODE_INITIAL:
 630                                 if ( $elm === self::NS_RDF . ' Description' ) {
 631                                         array_shift( $this->mode );
 632                                 } else {
 633                                         throw new MWException( 'Element ended unexpectedly while in MODE_INITIAL' );
 634                                 }
 635                                 break;
 636                         case self::MODE_LI:
 637                         case self::MODE_LI_LANG:
 638                                 $this->endElementModeLi( $elm );
 639                                 break;
 640                         case self::MODE_QDESC:
 641                                 $this->endElementModeQDesc( $elm );
 642                                 break;
 643                         default:
 644                                 wfDebugLog( 'XMP', __METHOD__ . " no mode (elm = $elm)" );
 645                                 break;
 646                 }
 647         }
 648
 649
 650         /**
 651         * Hit an opening element while in MODE_IGNORE
 652         *
 653         * XMP is extensible, so ignore any tag we don't understand.
 654         *
 655         * Mostly ignores, unless we encounter the element that we are ignoring.
 656         * in which case we add it to the item stack, so we can ignore things
 657         * that are nested, correctly.
 658         *
 659         * @param $elm String namespace . ' ' . tag name
 660         */
 661         private function startElementModeIgnore( $elm ) {
 662                 if ( $elm === $this->curItem[0] ) {
 663                         array_unshift( $this->curItem, $elm );
 664                         array_unshift( $this->mode, self::MODE_IGNORE );
 665                 }
 666         }
 667         /**
 668         *  Start element in MODE_BAG (unordered array)
 669         * this should always be <rdf:Bag>
 670         *
 671         * @param $elm String namespace . ' ' . tag
 672         * @throws MWException if we have an element that's not <rdf:Bag>
 673         */
 674         private function startElementModeBag( $elm ) {
 675                 if ( $elm === self::NS_RDF . ' Bag' ) {
 676                         array_unshift( $this->mode, self::MODE_LI );
 677                 } else {
 678                         throw new MWException( "Expected <rdf:Bag> but got $elm." );
 679                 }
 680
 681         }
 682         /**
 683         * Start element in MODE_SEQ (ordered array)
 684         * this should always be <rdf:Seq>
 685         *
 686         * @param $elm String namespace . ' ' . tag
 687         * @throws MWException if we have an element that's not <rdf:Seq>
 688         */
 689         private function startElementModeSeq( $elm ) {
 690                 if ( $elm === self::NS_RDF . ' Seq' ) {
 691                         array_unshift( $this->mode, self::MODE_LI );
 692                 } else if ( $elm === self::NS_RDF . ' Bag' ) {
 693                         # bug 27105
 694                         wfDebugLog( 'XMP', __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 695                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 696                         array_unshift( $this->mode, self::MODE_LI );
 697                 } else {
 698                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 699                 }
 700
 701         }
 702         /**
 703         * Start element in MODE_LANG (language alternative)
 704         * this should always be <rdf:Alt>
 705         *
 706         * This tag tends to be used for metadata like describe this
 707         * picture, which can be translated into multiple languages.
 708         *
 709         * XMP supports non-linguistic alternative selections,
 710         * which are really only used for thumbnails, which
 711         * we don't care about.
 712         *
 713         * @param $elm String namespace . ' ' . tag
 714         * @throws MWException if we have an element that's not <rdf:Alt>
 715         */
 716         private function startElementModeLang( $elm ) {
 717                 if ( $elm === self::NS_RDF . ' Alt' ) {
 718                         array_unshift( $this->mode, self::MODE_LI_LANG );
 719                 } else {
 720                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 721                 }
 722
 723         }
 724         /**
 725         * Handle an opening element when in MODE_SIMPLE
 726         *
 727         * This should not happen often. This is for if a simple element
 728         * already opened has a child element. Could happen for a
 729         * qualified element.
 730         *
 731         * For example:
 732         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 733         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 734         *   </exif:DigitalZoomRatio>
 735         *
 736         * This method is called when processing the <rdf:Description> element
 737         *
 738         * @param $elm String namespace and tag names separated by space.
 739         * @param $attribs Array Attributes of the element.
 740         */
 741         private function startElementModeSimple( $elm, $attribs ) {
 742                 if ( $elm === self::NS_RDF . ' Description' ) {
 743                         // If this value has qualifiers
 744                         array_unshift( $this->mode, self::MODE_QDESC );
 745                         array_unshift( $this->curItem, $this->curItem[0] );
 746
 747                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 748                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 749                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 750                         }
 751                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 752                         // This should not be here.
 753                         throw new MWException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 754
 755                 } else {
 756                         // something else we don't recognize, like a qualifier maybe.
 757                         wfDebugLog( 'XMP', __METHOD__ . " Encountered element <$elm> where only expecting character data as value of " . $this->curItem[0] );
 758                         array_unshift( $this->mode, self::MODE_IGNORE );
 759                         array_unshift( $this->curItem, $elm );
 760
 761                 }
 762
 763         }
 764         /**
 765         * Start an element when in MODE_QDESC.
 766         * This generally happens when a simple element has an inner
 767         * rdf:Description to hold qualifier elements.
 768         *
 769         * For example in:
 770         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 771         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 772         *   </exif:DigitalZoomRatio>
 773         * Called when processing the <rdf:value> or <foo:someQualifier>.
 774         *
 775         * @param $elm String namespace and tag name separated by a space.
 776         *
 777         */
 778         private function startElementModeQDesc( $elm ) {
 779                 if ( $elm === self::NS_RDF . ' value' ) {
 780                         return; // do nothing
 781                 } else {
 782                         // otherwise its a qualifier, which we ignore
 783                         array_unshift( $this->mode, self::MODE_IGNORE );
 784                         array_unshift( $this->curItem, $elm );
 785                 }
 786         }
 787         /**
 788         * Starting an element when in MODE_INITIAL
 789         * This usually happens when we hit an element inside
 790         * the outer rdf:Description
 791         *
 792         * This is generally where most properties start.
 793         *
 794         * @param $ns String Namespace
 795         * @param $tag String tag name (without namespace prefix)
 796         * @param $attribs Array array of attributes
 797         */
 798         private function startElementModeInitial( $ns, $tag, $attribs ) {
 799                 if ( $ns !== self::NS_RDF ) {
 800
 801                         if ( isset( $this->items[$ns][$tag] ) ) {
 802                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
 803                                         // If this element is supposed to appear only as
 804                                         // a child of a structure, but appears here (not as
 805                                         // a child of a struct), then something weird is
 806                                         // happening, so ignore this element and its children.
 807
 808                                         wfDebugLog( 'XMP', "Encountered <$ns:$tag> outside"
 809                                                 . " of its expected parent. Ignoring." );
 810
 811                                         array_unshift( $this->mode, self::MODE_IGNORE );
 812                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
 813                                         return;
 814                                 }
 815                                 $mode = $this->items[$ns][$tag]['mode'];
 816                                 array_unshift( $this->mode, $mode );
 817                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 818                                 if ( $mode === self::MODE_STRUCT ) {
 819                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
 820                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
 821                                 }
 822                                 if ( $this->charContent !== false ) {
 823                                         // Something weird.
 824                                         // Should not happen in valid XMP.
 825                                         throw new MWException( 'tag nested in non-whitespace characters.' );
 826                                 }
 827                         } else {
 828                                 // This element is not on our list of allowed elements so ignore.
 829                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
 830                                 array_unshift( $this->mode, self::MODE_IGNORE );
 831                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 832                                 return;
 833                         }
 834
 835                 }
 836                 // process attributes
 837                 $this->doAttribs( $attribs );
 838         }
 839         /**
 840         * Hit an opening element when in a Struct (MODE_STRUCT)
 841         * This is generally for fields of a compound property.
 842         *
 843         * Example of a struct (abbreviated; flash has more properties):
 844         *
 845         * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
 846         *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
 847         *
 848         * or:
 849         *
 850         * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
 851         *  <exif:Mode>1</exif:Mode></exif:Flash>
 852         *
 853         * @param $ns String namespace
 854         * @param $tag String tag name (no ns)
 855         * @param $attribs Array array of attribs w/ values.
 856         */
 857         private function startElementModeStruct( $ns, $tag, $attribs ) {
 858                 if ( $ns !== self::NS_RDF ) {
 859
 860                         if ( isset( $this->items[$ns][$tag] ) ) {
 861                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
 862                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] ) )
 863                                 {
 864                                         // This assumes that we don't have inter-namespace nesting
 865                                         // which we don't in all the properties we're interested in.
 866                                         throw new MWException( " <$tag> appeared nested in <" . $this->ancestorStruct
 867                                                 . "> where it is not allowed." );
 868                                 }
 869                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
 870                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 871                                 if ( $this->charContent !== false ) {
 872                                         // Something weird.
 873                                         // Should not happen in valid XMP.
 874                                         throw new MWException( "tag <$tag> nested in non-whitespace characters (" . $this->charContent . ")." );
 875                                 }
 876                         } else {
 877                                 array_unshift( $this->mode, self::MODE_IGNORE );
 878                                 array_unshift( $this->curItem, $elm );
 879                                 return;
 880                         }
 881
 882                 }
 883
 884                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
 885                         $this->doAttribs( $attribs );
 886                         array_unshift( $this->mode, self::MODE_STRUCT );
 887                         array_unshift( $this->curItem, $this->curItem[0] );
 888                 }
 889         }
 890         /**
 891         * opening element in MODE_LI
 892         * process elements of arrays.
 893         *
 894         * Example:
 895         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 896         *   </rdf:Seq> </exif:ISOSpeedRatings>
 897         * This method is called when we hit the <rdf:li> element.
 898         *
 899         * @param $elm String: namespace . ' ' . tagname
 900         * @param $attribs Array: Attributes. (needed for BAGSTRUCTS)
 901         * @throws MWException if gets a tag other than <rdf:li>
 902         */
 903         private function startElementModeLi( $elm, $attribs ) {
 904                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
 905                         throw new MWException( "<rdf:li> expected but got $elm." );
 906                 }
 907
 908                 if ( !isset( $this->mode[1] ) ) {
 909                         // This should never ever ever happen. Checking for it
 910                         // to be paranoid.
 911                         throw new MWException( 'In mode Li, but no 2xPrevious mode!' );
 912                 }
 913
 914                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
 915                         // This list item contains a compound (STRUCT) value.
 916                         array_unshift( $this->mode, self::MODE_STRUCT );
 917                         array_unshift( $this->curItem, $elm );
 918                         $this->processingArray = true;
 919
 920                         if ( !isset( $this->curItem[1] ) ) {
 921                                 // be paranoid.
 922                                 throw new MWException( 'Can not find parent of BAGSTRUCT.' );
 923                         }
 924                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
 925                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
 926                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
 927
 928                         $this->doAttribs( $attribs );
 929
 930                 } else {
 931                         // Normal BAG or SEQ containing simple values.
 932                         array_unshift( $this->mode, self::MODE_SIMPLE );
 933                         // need to add curItem[0] on again since one is for the specific item
 934                         // and one is for the entire group.
 935                         array_unshift( $this->curItem, $this->curItem[0] );
 936                         $this->processingArray = true;
 937                 }
 938
 939         }
 940         /**
 941         * Opening element in MODE_LI_LANG.
 942         * process elements of language alternatives
 943         *
 944         * Example:
 945         * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
 946         *  </rdf:li> </rdf:Alt> </dc:title>
 947         *
 948         * This method is called when we hit the <rdf:li> element.
 949         *
 950         * @param $elm String namespace . ' ' . tag
 951         * @param $attribs array array of elements (most importantly xml:lang)
 952         * @throws MWException if gets a tag other than <rdf:li> or if no xml:lang
 953         */
 954         private function startElementModeLiLang( $elm, $attribs ) {
 955                 if ( $elm !== self::NS_RDF . ' li' ) {
 956                         throw new MWException( __METHOD__ . " <rdf:li> expected but got $elm." );
 957                 }
 958                 if ( !isset( $attribs[ self::NS_XML . ' lang'] )
 959                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[ self::NS_XML . ' lang' ] ) )
 960                 {
 961                         throw new MWException( __METHOD__
 962                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
 963                 }
 964
 965                 // Lang is case-insensitive.
 966                 $this->itemLang = strtolower( $attribs[ self::NS_XML . ' lang' ] );
 967
 968                 // need to add curItem[0] on again since one is for the specific item
 969                 // and one is for the entire group.
 970                 array_unshift( $this->curItem, $this->curItem[0] );
 971                 array_unshift( $this->mode, self::MODE_SIMPLE );
 972                 $this->processingArray = true;
 973         }
 974
 975         /**
 976         * Hits an opening element.
 977         * Generally just calls a helper based on what MODE we're in.
 978         * Also does some initial set up for the wrapper element
 979         *
 980         * @param $parser XMLParser
 981         * @param $elm String namespace <space> element
 982         * @param $attribs Array attribute name => value
 983         */
 984         function startElement( $parser, $elm, $attribs ) {
 985
 986                 if ( $elm === self::NS_RDF . ' RDF'
 987                         || $elm === 'adobe:ns:meta/ xmpmeta'
 988                         || $elm === 'adobe:ns:meta/ xapmeta')
 989                 {
 990                         /* ignore. */
 991                         return;
 992                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
 993                         if ( count( $this->mode ) === 0 ) {
 994                                 // outer rdf:desc
 995                                 array_unshift( $this->mode, self::MODE_INITIAL );
 996                         }
 997                 } elseif ( $elm === self::NS_RDF . ' type' ) {
 998                         // This doesn't support rdf:type properly.
 999                         // In practise I have yet to see a file that
1000                         // uses this element, however it is mentioned
1001                         // on page 25 of part 1 of the xmp standard.
1002                         //
1003                         // also it seems as if exiv2 and exiftool do not support
1004                         // this either (That or I misunderstand the standard)
1005                         wfDebugLog( 'XMP', __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1006                 }
1007
1008                 if ( strpos( $elm, ' ' ) === false ) {
1009                         // This probably shouldn't happen.
1010                         wfDebugLog( 'XMP', __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1011                         return;
1012                 }
1013
1014                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1015
1016                 if ( count( $this->mode ) === 0 ) {
1017                         // This should not happen.
1018                         throw new MWException('Error extracting XMP, '
1019                                 . "encountered <$elm> with no mode" );
1020                 }
1021
1022                 switch( $this->mode[0] ) {
1023                         case self::MODE_IGNORE:
1024                                 $this->startElementModeIgnore( $elm );
1025                                 break;
1026                         case self::MODE_SIMPLE:
1027                                 $this->startElementModeSimple( $elm, $attribs );
1028                                 break;
1029                         case self::MODE_INITIAL:
1030                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1031                                 break;
1032                         case self::MODE_STRUCT:
1033                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1034                                 break;
1035                         case self::MODE_BAG:
1036                         case self::MODE_BAGSTRUCT:
1037                                 $this->startElementModeBag( $elm );
1038                                 break;
1039                         case self::MODE_SEQ:
1040                                 $this->startElementModeSeq( $elm );
1041                                 break;
1042                         case self::MODE_LANG:
1043                                 $this->startElementModeLang( $elm );
1044                                 break;
1045                         case self::MODE_LI_LANG:
1046                                 $this->startElementModeLiLang( $elm, $attribs );
1047                                 break;
1048                         case self::MODE_LI:
1049                                 $this->startElementModeLi( $elm, $attribs );
1050                                 break;
1051                         case self::MODE_QDESC:
1052                                 $this->startElementModeQDesc( $elm );
1053                                 break;
1054                         default:
1055                                 throw new MWException( 'StartElement in unknown mode: ' . $this->mode[0] );
1056                                 break;
1057                 }
1058
1059
1060
1061         }
1062         /**
1063         * Process attributes.
1064         * Simple values can be stored as either a tag or attribute
1065         *
1066         * Often the initial <rdf:Description> tag just has all the simple
1067         * properties as attributes.
1068         *
1069         * Example:
1070         * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1071         *
1072         * @param $attribs Array attribute=>value array.
1073         */
1074         private function doAttribs( $attribs ) {
1075
1076                 // first check for rdf:parseType attribute, as that can change
1077                 // how the attributes are interperted.
1078
1079                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1080                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1081                         && $this->mode[0] === self::MODE_SIMPLE )
1082                 {
1083                         // this is equivalent to having an inner rdf:Description
1084                         $this->mode[0] = self::MODE_QDESC;
1085                 }
1086                 foreach ( $attribs as $name => $val ) {
1087
1088
1089                         if ( strpos( $name, ' ' ) === false ) {
1090                                 // This shouldn't happen, but so far some old software forgets namespace
1091                                 // on rdf:about.
1092                                 wfDebugLog( 'XMP', __METHOD__ . ' Encountered non-namespaced attribute: '
1093                                         . " $name=\"$val\". Skipping. " );
1094                                 continue;
1095                         }
1096                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1097                         if ( $ns === self::NS_RDF ) {
1098                                 if ( $tag === 'value' || $tag === 'resource' ) {
1099                                         // resource is for url.
1100                                         // value attribute is a weird way of just putting the contents.
1101                                         $this->char( $this->xmlParser, $val );
1102                                 }
1103                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1104                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1105                                         throw new MWException( __METHOD__
1106                                                 . " $ns:$tag found as attribute where not allowed" );
1107                                 }
1108                                 $this->saveValue( $ns, $tag, $val );
1109                         } else {
1110                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1111                         }
1112                 }
1113         }
1114         /**
1115         * Given an extracted value, save it to results array
1116         *
1117         * note also uses $this->ancestorStruct and
1118         * $this->processingArray to determine what name to
1119         * save the value under. (in addition to $tag).
1120         *
1121         * @param $ns String namespace of tag this is for
1122         * @param $tag String tag name
1123         * @param $val String value to save
1124         */
1125         private function saveValue( $ns, $tag, $val ) {
1126
1127                 $info =& $this->items[$ns][$tag];
1128                 $finalName = isset( $info['map_name'] )
1129                         ? $info['map_name'] : $tag;
1130                 if ( isset( $info['validate'] ) ) {
1131                         $validate = is_array( $info['validate'] ) ? $info['validate']
1132                                 : array( 'XMPValidate', $info['validate'] );
1133
1134                         if ( is_callable( $validate ) ) {
1135                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1136                                 // the reasoning behind using &$val instead of using the return value
1137                                 // is to be consistent between here and validating structures.
1138                                 if ( is_null( $val ) ) {
1139                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
1140                                         return;
1141                                 }
1142                         } else {
1143                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
1144                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1145                         }
1146                 }
1147
1148                 if ( $this->ancestorStruct && $this->processingArray ) {
1149                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1150                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1151                 } elseif ( $this->ancestorStruct ) {
1152                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1153                 } elseif ( $this->processingArray ) {
1154                         if ( $this->itemLang === false ) {
1155                                 // normal array
1156                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1157                         } else {
1158                                 // lang array.
1159                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1160                         }
1161                 } else {
1162                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1163                 }
1164         }
1165
1166
1167 }