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         /** Destroy the xml parser
 112         *
 113         * Not sure if this is actually needed.
 114         */
 115         function __destruct() {
 116                 // not sure if this is needed.
 117                 xml_parser_free( $this->xmlParser );
 118         }
 119
 120         /** Get the result array. Do some post-processing before returning
 121         * the array, and transform any metadata that is special-cased.
 122         *
 123         * @return Array array of results as an array of arrays suitable for
 124         *       FormatMetadata::getFormattedData().
 125         */
 126         public function getResults() {
 127                 // xmp-special is for metadata that affects how stuff
 128                 // is extracted. For example xmpNote:HasExtendedXMP.
 129
 130                 // It is also used to handle photoshop:AuthorsPosition
 131                 // which is weird and really part of another property,
 132                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 133                 // The location fields also use it.
 134
 135                 $data = $this->results;
 136
 137                 wfRunHooks('XMPGetResults', Array(&$data));
 138
 139                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 140                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 141                         && isset( $data['xmp-general']['Artist'][0] )
 142                 ) {
 143                         // Note, if there is more than one creator,
 144                         // this only applies to first. This also will
 145                         // only apply to the dc:Creator prop, not the
 146                         // exif:Artist prop.
 147
 148                         $data['xmp-general']['Artist'][0] =
 149                                 $data['xmp-special']['AuthorsPosition'] . ', '
 150                                 . $data['xmp-general']['Artist'][0];
 151                 }
 152
 153                 // Go through the LocationShown and LocationCreated
 154                 // changing it to the non-hierarchal form used by
 155                 // the other location fields.
 156
 157                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 158                         && is_array( $data['xmp-special']['LocationShown'][0] )
 159                 ) {
 160                         // the is_array is just paranoia. It should always
 161                         // be an array.
 162                         foreach( $data['xmp-special']['LocationShown'] as $loc ) {
 163                                 if ( !is_array( $loc ) ) {
 164                                         // To avoid copying over the _type meta-fields.
 165                                         continue;
 166                                 }
 167                                 foreach( $loc as $field => $val ) {
 168                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 169                                 }
 170                         }
 171                 }
 172                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 173                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 174                 ) {
 175                         // the is_array is just paranoia. It should always
 176                         // be an array.
 177                         foreach( $data['xmp-special']['LocationCreated'] as $loc ) {
 178                                 if ( !is_array( $loc ) ) {
 179                                         // To avoid copying over the _type meta-fields.
 180                                         continue;
 181                                 }
 182                                 foreach(  $loc as $field => $val ) {
 183                                         $data['xmp-general'][$field . 'Created'][] = $val;
 184                                 }
 185                         }
 186                 }
 187
 188
 189                 // We don't want to return the special values, since they're
 190                 // special and not info to be stored about the file.
 191                 unset( $data['xmp-special'] );
 192
 193                 // Convert GPSAltitude to negative if below sea level.
 194                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] ) ) {
 195                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1'
 196                                 && isset( $data['xmp-exif']['GPSAltitude'] )
 197                         ) {
 198                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 199                         }
 200                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 201                 }
 202
 203                 return $data;
 204         }
 205
 206         /**
 207         * Main function to call to parse XMP. Use getResults to
 208         * get results.
 209         *
 210         * Also catches any errors during processing, writes them to
 211         * debug log, blanks result array and returns false.
 212         *
 213         * @param String: $content XMP data
 214         * @param Boolean: $allOfIt If this is all the data (true) or if its split up (false). Default true
 215         * @param Boolean: $reset - does xml parser need to be reset. Default false
 216         * @return Boolean success.
 217         */
 218         public function parse( $content, $allOfIt = true, $reset = false ) {
 219                 if ( $reset ) {
 220                         $this->resetXMLParser();
 221                 }
 222                 try {
 223
 224                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 225                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 226                         if ( !$this->charset ) {
 227                                 $bom = array();
 228                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 229                                          $content, $bom )
 230                                 ) {
 231                                         switch ( $bom[0] ) {
 232                                                 case "\xFE\xFF":
 233                                                         $this->charset = 'UTF-16BE';
 234                                                         break;
 235                                                 case "\xFF\xFE":
 236                                                         $this->charset = 'UTF-16LE';
 237                                                         break;
 238                                                 case "\x00\x00\xFE\xFF":
 239                                                         $this->charset = 'UTF-32BE';
 240                                                         break;
 241                                                 case "\xFF\xFE\x00\x00":
 242                                                         $this->charset = 'UTF-32LE';
 243                                                         break;
 244                                                 case "\xEF\xBB\xBF":
 245                                                         $this->charset = 'UTF-8';
 246                                                         break;
 247                                                 default:
 248                                                         //this should be impossible to get to
 249                                                         throw new MWException("Invalid BOM");
 250                                                         break;
 251
 252                                         }
 253
 254                                 } else {
 255                                         // standard specifically says, if no bom assume utf-8
 256                                         $this->charset = 'UTF-8';
 257                                 }
 258                         }
 259                         if ( $this->charset !== 'UTF-8' ) {
 260                                 //don't convert if already utf-8
 261                                 wfSuppressWarnings();
 262                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 263                                 wfRestoreWarnings();
 264                         }
 265
 266                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 267                         if ( !$ok ) {
 268                                 $error = xml_error_string( xml_get_error_code( $this->xmlParser ) );
 269                                 $where = 'line: ' . xml_get_current_line_number( $this->xmlParser )
 270                                         . ' column: ' . xml_get_current_column_number( $this->xmlParser )
 271                                         . ' byte offset: ' . xml_get_current_byte_index( $this->xmlParser );
 272
 273                                 wfDebugLog( 'XMP', "XMPReader::parse : Error reading XMP content: $error ($where)" );
 274                                 $this->results = array(); // blank if error.
 275                                 return false;
 276                         }
 277                 } catch ( MWException $e ) {
 278                         wfDebugLog( 'XMP', 'XMP parse error: ' . $e );
 279                         $this->results = array();
 280                         return false;
 281                 }
 282                 return true;
 283         }
 284
 285         /** Entry point for XMPExtended blocks in jpeg files
 286          *
 287          * @todo In serious need of testing
 288          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 289          * @param String $content XMPExtended block minus the namespace signature
 290          * @return Boolean If it succeeded.
 291          */
 292         public function parseExtended( $content ) {
 293                 // @todo FIXME: This is untested. Hard to find example files
 294                 // or programs that make such files..
 295                 $guid = substr( $content, 0, 32 );
 296                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 297                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid ) {
 298                         wfDebugLog('XMP', __METHOD__ . " Ignoring XMPExtended block due to wrong guid (guid= '$guid' )");
 299                         return false;
 300                 }
 301                 $len  = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 302
 303                 if (!$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 304                         wfDebugLog('XMP', __METHOD__ . 'Error reading extended XMP block, invalid length or offset.');
 305                         return false;
 306                 }
 307
 308
 309                 // we're not very robust here. we should accept it in the wrong order. To quote
 310                 // the xmp standard:
 311                 // "A JPEG writer should write the ExtendedXMP marker segments in order, immediately following the
 312                 // StandardXMP. However, the JPEG standard does not require preservation of marker segment order. A
 313                 // robust JPEG reader should tolerate the marker segments in any order."
 314                 //
 315                 // otoh the probability that an image will have more than 128k of metadata is rather low...
 316                 // so the probability that it will have > 128k, and be in the wrong order is very low...
 317
 318                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 319                         wfDebugLog('XMP', __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 320                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')');
 321                         return false;
 322                 }
 323
 324                 if ( $len['offset'] === 0 ) {
 325                         // if we're starting the extended block, we've probably already
 326                         // done the XMPStandard block, so reset.
 327                         $this->resetXMLParser();
 328                 }
 329
 330                 $this->extendedXMPOffset += $len['length'];
 331
 332                 $actualContent = substr( $content, 40 );
 333
 334                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 335                         $atEnd = true;
 336                 } else {
 337                         $atEnd = false;
 338                 }
 339
 340                 wfDebugLog('XMP', __METHOD__ . 'Parsing a XMPExtended block');
 341                 return $this->parse( $actualContent, $atEnd );
 342         }
 343
 344         /**
 345         * Character data handler
 346         * Called whenever character data is found in the xmp document.
 347         *
 348         * does nothing if we're in MODE_IGNORE or if the data is whitespace
 349         * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 350         * data in the other modes).
 351         *
 352         * As an example, this happens when we encounter XMP like:
 353         * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 354         * and are processing the 0/10 bit.
 355         *
 356         * @param $parser XMLParser reference to the xml parser
 357         * @param $data String Character data
 358         * @throws MWException on invalid data
 359         */
 360         function char( $parser, $data ) {
 361
 362                 $data = trim( $data );
 363                 if ( trim( $data ) === "" ) {
 364                         return;
 365                 }
 366
 367                 if ( !isset( $this->mode[0] ) ) {
 368                         throw new MWException( 'Unexpected character data before first rdf:Description element' );
 369                 }
 370
 371                 if ( $this->mode[0] === self::MODE_IGNORE ) return;
 372
 373                 if ( $this->mode[0] !== self::MODE_SIMPLE
 374                         && $this->mode[0] !== self::MODE_QDESC
 375                 ) {
 376                         throw new MWException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 377                 }
 378
 379                 // to check, how does this handle w.s.
 380                 if ( $this->charContent === false ) {
 381                         $this->charContent = $data;
 382                 } else {
 383                         $this->charContent .= $data;
 384                 }
 385
 386         }
 387
 388         /** When we hit a closing element in MODE_IGNORE
 389         * Check to see if this is the element we started to ignore,
 390         * in which case we get out of MODE_IGNORE
 391         *
 392         * @param $elm String Namespace of element followed by a space and then tag name of element.
 393         */
 394         private function endElementModeIgnore ( $elm ) {
 395
 396                 if ( $this->curItem[0] === $elm ) {
 397                         array_shift( $this->curItem );
 398                         array_shift( $this->mode );
 399                 }
 400                 return;
 401
 402         }
 403
 404         /**
 405         * Hit a closing element when in MODE_SIMPLE.
 406         * This generally means that we finished processing a
 407         * property value, and now have to save the result to the
 408         * results array
 409         *
 410         * For example, when processing:
 411         * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 412         * this deals with when we hit </exif:DigitalZoomRatio>.
 413         *
 414         * Or it could be if we hit the end element of a property
 415         * of a compound data structure (like a member of an array).
 416         *
 417         * @param $elm String namespace, space, and tag name.
 418         */
 419         private function endElementModeSimple ( $elm ) {
 420                 if ( $this->charContent !== false ) {
 421                         if ( $this->processingArray ) {
 422                                 // if we're processing an array, use the original element
 423                                 // name instead of rdf:li.
 424                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 425                         } else {
 426                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 427                         }
 428                         $this->saveValue( $ns, $tag, $this->charContent );
 429
 430                         $this->charContent = false; // reset
 431                 }
 432                 array_shift( $this->curItem );
 433                 array_shift( $this->mode );
 434
 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         /**
 546         * End element while in MODE_QDESC
 547         * mostly when ending an element when we have a simple value
 548         * that has qualifiers.
 549         *
 550         * Qualifiers aren't all that common, and we don't do anything
 551         * with them.
 552         *
 553         * @param $elm String namespace and element
 554         */
 555         private function endElementModeQDesc( $elm ) {
 556
 557                 if ( $elm === self::NS_RDF . ' value' ) {
 558                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 559                         $this->saveValue( $ns, $tag, $this->charContent );
 560                         return;
 561                 } else {
 562                         array_shift( $this->mode );
 563                         array_shift( $this->curItem );
 564                 }
 565
 566
 567         }
 568
 569         /**
 570         * Handler for hitting a closing element.
 571         *
 572         * generally just calls a helper function depending on what
 573         * mode we're in.
 574         *
 575         * Ignores the outer wrapping elements that are optional in
 576         * xmp and have no meaning.
 577         *
 578         * @param $parser XMLParser
 579         * @param $elm String namespace . ' ' . element name
 580         */
 581         function endElement( $parser, $elm ) {
 582                 if ( $elm === ( self::NS_RDF . ' RDF' )
 583                         || $elm === 'adobe:ns:meta/ xmpmeta'
 584                         || $elm === 'adobe:ns:meta/ xapmeta' )
 585                 {
 586                         // ignore these.
 587                         return;
 588                 }
 589
 590                 if ( $elm === self::NS_RDF . ' type' ) {
 591                         // these aren't really supported properly yet.
 592                         // However, it appears they almost never used.
 593                         wfDebugLog( 'XMP', __METHOD__ . ' encountered <rdf:type>' );
 594                 }
 595
 596                 if ( strpos( $elm, ' ' ) === false ) {
 597                         // This probably shouldn't happen.
 598                         // However, there is a bug in an adobe product
 599                         // that forgets the namespace on some things.
 600                         // (Luckily they are unimportant things).
 601                         wfDebugLog( 'XMP', __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 602                         return;
 603                 }
 604
 605                 if ( count( $this->mode[0] ) === 0 ) {
 606                         // This should never ever happen and means
 607                         // there is a pretty major bug in this class.
 608                         throw new MWException( 'Encountered end element with no mode' );
 609                 }
 610
 611                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 612                         // just to be paranoid. Should always have a curItem, except for initially
 613                         // (aka during MODE_INITAL).
 614                         throw new MWException( "Hit end element </$elm> but no curItem" );
 615                 }
 616
 617                 switch( $this->mode[0] ) {
 618                         case self::MODE_IGNORE:
 619                                 $this->endElementModeIgnore( $elm );
 620                                 break;
 621                         case self::MODE_SIMPLE:
 622                                 $this->endElementModeSimple( $elm );
 623                                 break;
 624                         case self::MODE_STRUCT:
 625                         case self::MODE_SEQ:
 626                         case self::MODE_BAG:
 627                         case self::MODE_LANG:
 628                         case self::MODE_BAGSTRUCT:
 629                                 $this->endElementNested( $elm );
 630                                 break;
 631                         case self::MODE_INITIAL:
 632                                 if ( $elm === self::NS_RDF . ' Description' ) {
 633                                         array_shift( $this->mode );
 634                                 } else {
 635                                         throw new MWException( 'Element ended unexpectedly while in MODE_INITIAL' );
 636                                 }
 637                                 break;
 638                         case self::MODE_LI:
 639                         case self::MODE_LI_LANG:
 640                                 $this->endElementModeLi( $elm );
 641                                 break;
 642                         case self::MODE_QDESC:
 643                                 $this->endElementModeQDesc( $elm );
 644                                 break;
 645                         default:
 646                                 wfDebugLog( 'XMP', __METHOD__ . " no mode (elm = $elm)" );
 647                                 break;
 648                 }
 649         }
 650
 651         /**
 652         * Hit an opening element while in MODE_IGNORE
 653         *
 654         * XMP is extensible, so ignore any tag we don't understand.
 655         *
 656         * Mostly ignores, unless we encounter the element that we are ignoring.
 657         * in which case we add it to the item stack, so we can ignore things
 658         * that are nested, correctly.
 659         *
 660         * @param $elm String namespace . ' ' . tag name
 661         */
 662         private function startElementModeIgnore( $elm ) {
 663                 if ( $elm === $this->curItem[0] ) {
 664                         array_unshift( $this->curItem, $elm );
 665                         array_unshift( $this->mode, self::MODE_IGNORE );
 666                 }
 667         }
 668
 669         /**
 670         *  Start element in MODE_BAG (unordered array)
 671         * this should always be <rdf:Bag>
 672         *
 673         * @param $elm String namespace . ' ' . tag
 674         * @throws MWException if we have an element that's not <rdf:Bag>
 675         */
 676         private function startElementModeBag( $elm ) {
 677                 if ( $elm === self::NS_RDF . ' Bag' ) {
 678                         array_unshift( $this->mode, self::MODE_LI );
 679                 } else {
 680                         throw new MWException( "Expected <rdf:Bag> but got $elm." );
 681                 }
 682
 683         }
 684
 685         /**
 686         * Start element in MODE_SEQ (ordered array)
 687         * this should always be <rdf:Seq>
 688         *
 689         * @param $elm String namespace . ' ' . tag
 690         * @throws MWException if we have an element that's not <rdf:Seq>
 691         */
 692         private function startElementModeSeq( $elm ) {
 693                 if ( $elm === self::NS_RDF . ' Seq' ) {
 694                         array_unshift( $this->mode, self::MODE_LI );
 695                 } else if ( $elm === self::NS_RDF . ' Bag' ) {
 696                         # bug 27105
 697                         wfDebugLog( 'XMP', __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 698                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 699                         array_unshift( $this->mode, self::MODE_LI );
 700                 } else {
 701                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 702                 }
 703
 704         }
 705
 706         /**
 707         * Start element in MODE_LANG (language alternative)
 708         * this should always be <rdf:Alt>
 709         *
 710         * This tag tends to be used for metadata like describe this
 711         * picture, which can be translated into multiple languages.
 712         *
 713         * XMP supports non-linguistic alternative selections,
 714         * which are really only used for thumbnails, which
 715         * we don't care about.
 716         *
 717         * @param $elm String namespace . ' ' . tag
 718         * @throws MWException if we have an element that's not <rdf:Alt>
 719         */
 720         private function startElementModeLang( $elm ) {
 721                 if ( $elm === self::NS_RDF . ' Alt' ) {
 722                         array_unshift( $this->mode, self::MODE_LI_LANG );
 723                 } else {
 724                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 725                 }
 726
 727         }
 728
 729         /**
 730         * Handle an opening element when in MODE_SIMPLE
 731         *
 732         * This should not happen often. This is for if a simple element
 733         * already opened has a child element. Could happen for a
 734         * qualified element.
 735         *
 736         * For example:
 737         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 738         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 739         *   </exif:DigitalZoomRatio>
 740         *
 741         * This method is called when processing the <rdf:Description> element
 742         *
 743         * @param $elm String namespace and tag names separated by space.
 744         * @param $attribs Array Attributes of the element.
 745         */
 746         private function startElementModeSimple( $elm, $attribs ) {
 747                 if ( $elm === self::NS_RDF . ' Description' ) {
 748                         // If this value has qualifiers
 749                         array_unshift( $this->mode, self::MODE_QDESC );
 750                         array_unshift( $this->curItem, $this->curItem[0] );
 751
 752                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 753                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 754                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 755                         }
 756                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 757                         // This should not be here.
 758                         throw new MWException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 759
 760                 } else {
 761                         // something else we don't recognize, like a qualifier maybe.
 762                         wfDebugLog( 'XMP', __METHOD__ . " Encountered element <$elm> where only expecting character data as value of " . $this->curItem[0] );
 763                         array_unshift( $this->mode, self::MODE_IGNORE );
 764                         array_unshift( $this->curItem, $elm );
 765
 766                 }
 767
 768         }
 769
 770         /**
 771         * Start an element when in MODE_QDESC.
 772         * This generally happens when a simple element has an inner
 773         * rdf:Description to hold qualifier elements.
 774         *
 775         * For example in:
 776         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 777         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 778         *   </exif:DigitalZoomRatio>
 779         * Called when processing the <rdf:value> or <foo:someQualifier>.
 780         *
 781         * @param $elm String namespace and tag name separated by a space.
 782         *
 783         */
 784         private function startElementModeQDesc( $elm ) {
 785                 if ( $elm === self::NS_RDF . ' value' ) {
 786                         return; // do nothing
 787                 } else {
 788                         // otherwise its a qualifier, which we ignore
 789                         array_unshift( $this->mode, self::MODE_IGNORE );
 790                         array_unshift( $this->curItem, $elm );
 791                 }
 792         }
 793
 794         /**
 795         * Starting an element when in MODE_INITIAL
 796         * This usually happens when we hit an element inside
 797         * the outer rdf:Description
 798         *
 799         * This is generally where most properties start.
 800         *
 801         * @param $ns String Namespace
 802         * @param $tag String tag name (without namespace prefix)
 803         * @param $attribs Array array of attributes
 804         */
 805         private function startElementModeInitial( $ns, $tag, $attribs ) {
 806                 if ( $ns !== self::NS_RDF ) {
 807
 808                         if ( isset( $this->items[$ns][$tag] ) ) {
 809                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
 810                                         // If this element is supposed to appear only as
 811                                         // a child of a structure, but appears here (not as
 812                                         // a child of a struct), then something weird is
 813                                         // happening, so ignore this element and its children.
 814
 815                                         wfDebugLog( 'XMP', "Encountered <$ns:$tag> outside"
 816                                                 . " of its expected parent. Ignoring." );
 817
 818                                         array_unshift( $this->mode, self::MODE_IGNORE );
 819                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
 820                                         return;
 821                                 }
 822                                 $mode = $this->items[$ns][$tag]['mode'];
 823                                 array_unshift( $this->mode, $mode );
 824                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 825                                 if ( $mode === self::MODE_STRUCT ) {
 826                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
 827                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
 828                                 }
 829                                 if ( $this->charContent !== false ) {
 830                                         // Something weird.
 831                                         // Should not happen in valid XMP.
 832                                         throw new MWException( 'tag nested in non-whitespace characters.' );
 833                                 }
 834                         } else {
 835                                 // This element is not on our list of allowed elements so ignore.
 836                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
 837                                 array_unshift( $this->mode, self::MODE_IGNORE );
 838                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 839                                 return;
 840                         }
 841
 842                 }
 843                 // process attributes
 844                 $this->doAttribs( $attribs );
 845         }
 846
 847         /**
 848         * Hit an opening element when in a Struct (MODE_STRUCT)
 849         * This is generally for fields of a compound property.
 850         *
 851         * Example of a struct (abbreviated; flash has more properties):
 852         *
 853         * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
 854         *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
 855         *
 856         * or:
 857         *
 858         * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
 859         *  <exif:Mode>1</exif:Mode></exif:Flash>
 860         *
 861         * @param $ns String namespace
 862         * @param $tag String tag name (no ns)
 863         * @param $attribs Array array of attribs w/ values.
 864         */
 865         private function startElementModeStruct( $ns, $tag, $attribs ) {
 866                 if ( $ns !== self::NS_RDF ) {
 867
 868                         if ( isset( $this->items[$ns][$tag] ) ) {
 869                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
 870                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] ) )
 871                                 {
 872                                         // This assumes that we don't have inter-namespace nesting
 873                                         // which we don't in all the properties we're interested in.
 874                                         throw new MWException( " <$tag> appeared nested in <" . $this->ancestorStruct
 875                                                 . "> where it is not allowed." );
 876                                 }
 877                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
 878                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 879                                 if ( $this->charContent !== false ) {
 880                                         // Something weird.
 881                                         // Should not happen in valid XMP.
 882                                         throw new MWException( "tag <$tag> nested in non-whitespace characters (" . $this->charContent . ")." );
 883                                 }
 884                         } else {
 885                                 array_unshift( $this->mode, self::MODE_IGNORE );
 886                                 array_unshift( $this->curItem, $elm );
 887                                 return;
 888                         }
 889
 890                 }
 891
 892                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
 893                         $this->doAttribs( $attribs );
 894                         array_unshift( $this->mode, self::MODE_STRUCT );
 895                         array_unshift( $this->curItem, $this->curItem[0] );
 896                 }
 897         }
 898
 899         /**
 900         * opening element in MODE_LI
 901         * process elements of arrays.
 902         *
 903         * Example:
 904         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 905         *   </rdf:Seq> </exif:ISOSpeedRatings>
 906         * This method is called when we hit the <rdf:li> element.
 907         *
 908         * @param $elm String: namespace . ' ' . tagname
 909         * @param $attribs Array: Attributes. (needed for BAGSTRUCTS)
 910         * @throws MWException if gets a tag other than <rdf:li>
 911         */
 912         private function startElementModeLi( $elm, $attribs ) {
 913                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
 914                         throw new MWException( "<rdf:li> expected but got $elm." );
 915                 }
 916
 917                 if ( !isset( $this->mode[1] ) ) {
 918                         // This should never ever ever happen. Checking for it
 919                         // to be paranoid.
 920                         throw new MWException( 'In mode Li, but no 2xPrevious mode!' );
 921                 }
 922
 923                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
 924                         // This list item contains a compound (STRUCT) value.
 925                         array_unshift( $this->mode, self::MODE_STRUCT );
 926                         array_unshift( $this->curItem, $elm );
 927                         $this->processingArray = true;
 928
 929                         if ( !isset( $this->curItem[1] ) ) {
 930                                 // be paranoid.
 931                                 throw new MWException( 'Can not find parent of BAGSTRUCT.' );
 932                         }
 933                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
 934                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
 935                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
 936
 937                         $this->doAttribs( $attribs );
 938
 939                 } else {
 940                         // Normal BAG or SEQ containing simple values.
 941                         array_unshift( $this->mode, self::MODE_SIMPLE );
 942                         // need to add curItem[0] on again since one is for the specific item
 943                         // and one is for the entire group.
 944                         array_unshift( $this->curItem, $this->curItem[0] );
 945                         $this->processingArray = true;
 946                 }
 947
 948         }
 949
 950         /**
 951         * Opening element in MODE_LI_LANG.
 952         * process elements of language alternatives
 953         *
 954         * Example:
 955         * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
 956         *  </rdf:li> </rdf:Alt> </dc:title>
 957         *
 958         * This method is called when we hit the <rdf:li> element.
 959         *
 960         * @param $elm String namespace . ' ' . tag
 961         * @param $attribs array array of elements (most importantly xml:lang)
 962         * @throws MWException if gets a tag other than <rdf:li> or if no xml:lang
 963         */
 964         private function startElementModeLiLang( $elm, $attribs ) {
 965                 if ( $elm !== self::NS_RDF . ' li' ) {
 966                         throw new MWException( __METHOD__ . " <rdf:li> expected but got $elm." );
 967                 }
 968                 if ( !isset( $attribs[ self::NS_XML . ' lang'] )
 969                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[ self::NS_XML . ' lang' ] ) )
 970                 {
 971                         throw new MWException( __METHOD__
 972                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
 973                 }
 974
 975                 // Lang is case-insensitive.
 976                 $this->itemLang = strtolower( $attribs[ self::NS_XML . ' lang' ] );
 977
 978                 // need to add curItem[0] on again since one is for the specific item
 979                 // and one is for the entire group.
 980                 array_unshift( $this->curItem, $this->curItem[0] );
 981                 array_unshift( $this->mode, self::MODE_SIMPLE );
 982                 $this->processingArray = true;
 983         }
 984
 985         /**
 986         * Hits an opening element.
 987         * Generally just calls a helper based on what MODE we're in.
 988         * Also does some initial set up for the wrapper element
 989         *
 990         * @param $parser XMLParser
 991         * @param $elm String namespace <space> element
 992         * @param $attribs Array attribute name => value
 993         */
 994         function startElement( $parser, $elm, $attribs ) {
 995
 996                 if ( $elm === self::NS_RDF . ' RDF'
 997                         || $elm === 'adobe:ns:meta/ xmpmeta'
 998                         || $elm === 'adobe:ns:meta/ xapmeta')
 999                 {
1000                         /* ignore. */
1001                         return;
1002                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1003                         if ( count( $this->mode ) === 0 ) {
1004                                 // outer rdf:desc
1005                                 array_unshift( $this->mode, self::MODE_INITIAL );
1006                         }
1007                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1008                         // This doesn't support rdf:type properly.
1009                         // In practise I have yet to see a file that
1010                         // uses this element, however it is mentioned
1011                         // on page 25 of part 1 of the xmp standard.
1012                         //
1013                         // also it seems as if exiv2 and exiftool do not support
1014                         // this either (That or I misunderstand the standard)
1015                         wfDebugLog( 'XMP', __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1016                 }
1017
1018                 if ( strpos( $elm, ' ' ) === false ) {
1019                         // This probably shouldn't happen.
1020                         wfDebugLog( 'XMP', __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1021                         return;
1022                 }
1023
1024                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1025
1026                 if ( count( $this->mode ) === 0 ) {
1027                         // This should not happen.
1028                         throw new MWException('Error extracting XMP, '
1029                                 . "encountered <$elm> with no mode" );
1030                 }
1031
1032                 switch( $this->mode[0] ) {
1033                         case self::MODE_IGNORE:
1034                                 $this->startElementModeIgnore( $elm );
1035                                 break;
1036                         case self::MODE_SIMPLE:
1037                                 $this->startElementModeSimple( $elm, $attribs );
1038                                 break;
1039                         case self::MODE_INITIAL:
1040                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1041                                 break;
1042                         case self::MODE_STRUCT:
1043                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1044                                 break;
1045                         case self::MODE_BAG:
1046                         case self::MODE_BAGSTRUCT:
1047                                 $this->startElementModeBag( $elm );
1048                                 break;
1049                         case self::MODE_SEQ:
1050                                 $this->startElementModeSeq( $elm );
1051                                 break;
1052                         case self::MODE_LANG:
1053                                 $this->startElementModeLang( $elm );
1054                                 break;
1055                         case self::MODE_LI_LANG:
1056                                 $this->startElementModeLiLang( $elm, $attribs );
1057                                 break;
1058                         case self::MODE_LI:
1059                                 $this->startElementModeLi( $elm, $attribs );
1060                                 break;
1061                         case self::MODE_QDESC:
1062                                 $this->startElementModeQDesc( $elm );
1063                                 break;
1064                         default:
1065                                 throw new MWException( 'StartElement in unknown mode: ' . $this->mode[0] );
1066                                 break;
1067                 }
1068         }
1069
1070         /**
1071         * Process attributes.
1072         * Simple values can be stored as either a tag or attribute
1073         *
1074         * Often the initial <rdf:Description> tag just has all the simple
1075         * properties as attributes.
1076         *
1077         * Example:
1078         * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1079         *
1080         * @param $attribs Array attribute=>value array.
1081         */
1082         private function doAttribs( $attribs ) {
1083
1084                 // first check for rdf:parseType attribute, as that can change
1085                 // how the attributes are interperted.
1086
1087                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1088                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1089                         && $this->mode[0] === self::MODE_SIMPLE )
1090                 {
1091                         // this is equivalent to having an inner rdf:Description
1092                         $this->mode[0] = self::MODE_QDESC;
1093                 }
1094                 foreach ( $attribs as $name => $val ) {
1095
1096
1097                         if ( strpos( $name, ' ' ) === false ) {
1098                                 // This shouldn't happen, but so far some old software forgets namespace
1099                                 // on rdf:about.
1100                                 wfDebugLog( 'XMP', __METHOD__ . ' Encountered non-namespaced attribute: '
1101                                         . " $name=\"$val\". Skipping. " );
1102                                 continue;
1103                         }
1104                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1105                         if ( $ns === self::NS_RDF ) {
1106                                 if ( $tag === 'value' || $tag === 'resource' ) {
1107                                         // resource is for url.
1108                                         // value attribute is a weird way of just putting the contents.
1109                                         $this->char( $this->xmlParser, $val );
1110                                 }
1111                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1112                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1113                                         throw new MWException( __METHOD__
1114                                                 . " $ns:$tag found as attribute where not allowed" );
1115                                 }
1116                                 $this->saveValue( $ns, $tag, $val );
1117                         } else {
1118                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1119                         }
1120                 }
1121         }
1122
1123         /**
1124         * Given an extracted value, save it to results array
1125         *
1126         * note also uses $this->ancestorStruct and
1127         * $this->processingArray to determine what name to
1128         * save the value under. (in addition to $tag).
1129         *
1130         * @param $ns String namespace of tag this is for
1131         * @param $tag String tag name
1132         * @param $val String value to save
1133         */
1134         private function saveValue( $ns, $tag, $val ) {
1135
1136                 $info =& $this->items[$ns][$tag];
1137                 $finalName = isset( $info['map_name'] )
1138                         ? $info['map_name'] : $tag;
1139                 if ( isset( $info['validate'] ) ) {
1140                         $validate = is_array( $info['validate'] ) ? $info['validate']
1141                                 : array( 'XMPValidate', $info['validate'] );
1142
1143                         if ( is_callable( $validate ) ) {
1144                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1145                                 // the reasoning behind using &$val instead of using the return value
1146                                 // is to be consistent between here and validating structures.
1147                                 if ( is_null( $val ) ) {
1148                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
1149                                         return;
1150                                 }
1151                         } else {
1152                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
1153                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1154                         }
1155                 }
1156
1157                 if ( $this->ancestorStruct && $this->processingArray ) {
1158                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1159                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1160                 } elseif ( $this->ancestorStruct ) {
1161                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1162                 } elseif ( $this->processingArray ) {
1163                         if ( $this->itemLang === false ) {
1164                                 // normal array
1165                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1166                         } else {
1167                                 // lang array.
1168                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1169                         }
1170                 } else {
1171                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1172                 }
1173         }
1174 }