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'] !== $guid ) {
 300                         wfDebugLog('XMP', __METHOD__ . " Ignoring XMPExtended block due to wrong guid (guid= '$guid' )");
 301                         return false;
 302                 }
 303                 $len  = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 304
 305                 if (!$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 306                         wfDebugLog('XMP', __METHOD__ . 'Error reading extended XMP block, invalid length or offset.');
 307                         return false;
 308                 }
 309
 310
 311                 // we're not very robust here. we should accept it in the wrong order. To quote
 312                 // the xmp standard:
 313                 // "A JPEG writer should write the ExtendedXMP marker segments in order, immediately following the
 314                 // StandardXMP. However, the JPEG standard does not require preservation of marker segment order. A
 315                 // robust JPEG reader should tolerate the marker segments in any order."
 316                 //
 317                 // otoh the probability that an image will have more than 128k of metadata is rather low...
 318                 // so the probability that it will have > 128k, and be in the wrong order is very low...
 319
 320                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 321                         wfDebugLog('XMP', __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 322                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')');
 323                         return false;
 324                 }
 325
 326                 if ( $len['offset'] === 0 ) {
 327                         // if we're starting the extended block, we've probably already
 328                         // done the XMPStandard block, so reset.
 329                         $this->resetXMLParser();
 330                 }
 331
 332                 $this->extendedXMPOffset += $len['length'];
 333
 334                 $actualContent = substr( $content, 40 );
 335
 336                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 337                         $atEnd = true;
 338                 } else {
 339                         $atEnd = false;
 340                 }
 341
 342                 wfDebugLog('XMP', __METHOD__ . 'Parsing a XMPExtended block');
 343                 return $this->parse( $actualContent, $atEnd );
 344         }
 345
 346         /**
 347         * Character data handler
 348         * Called whenever character data is found in the xmp document.
 349         *
 350         * does nothing if we're in MODE_IGNORE or if the data is whitespace
 351         * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 352         * data in the other modes).
 353         *
 354         * As an example, this happens when we encounter XMP like:
 355         * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 356         * and are processing the 0/10 bit.
 357         *
 358         * @param $parser XMLParser reference to the xml parser
 359         * @param $data String Character data
 360         * @throws MWException on invalid data
 361         */
 362         function char( $parser, $data ) {
 363
 364                 $data = trim( $data );
 365                 if ( trim( $data ) === "" ) {
 366                         return;
 367                 }
 368
 369                 if ( !isset( $this->mode[0] ) ) {
 370                         throw new MWException( 'Unexpected character data before first rdf:Description element' );
 371                 }
 372
 373                 if ( $this->mode[0] === self::MODE_IGNORE ) return;
 374
 375                 if ( $this->mode[0] !== self::MODE_SIMPLE
 376                         && $this->mode[0] !== self::MODE_QDESC
 377                 ) {
 378                         throw new MWException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 379                 }
 380
 381                 // to check, how does this handle w.s.
 382                 if ( $this->charContent === false ) {
 383                         $this->charContent = $data;
 384                 } else {
 385                         $this->charContent .= $data;
 386                 }
 387
 388         }
 389         /** When we hit a closing element in MODE_IGNORE
 390         * Check to see if this is the element we started to ignore,
 391         * in which case we get out of MODE_IGNORE
 392         *
 393         * @param $elm String Namespace of element followed by a space and then tag name of element.
 394         */
 395         private function endElementModeIgnore ( $elm ) {
 396
 397                 if ( $this->curItem[0] === $elm ) {
 398                         array_shift( $this->curItem );
 399                         array_shift( $this->mode );
 400                 }
 401                 return;
 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         * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 438         * generally means we've finished processing a nested structure.
 439         * resets some internal variables to indicate that.
 440         *
 441         * Note this means we hit the </closing element> not the </rdf:Seq>.
 442         *
 443         * For example, when processing:
 444         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 445         *   </rdf:Seq> </exif:ISOSpeedRatings>
 446         *
 447         * This method is called when we hit the </exif:ISOSpeedRatings> tag.
 448         *
 449         * @param $elm String namespace . space . tag name.
 450         */
 451         private function endElementNested( $elm ) {
 452
 453                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 454                    in which case it could also be rdf:Description */
 455                 if ( $this->curItem[0] !== $elm
 456                         && !( $elm === self::NS_RDF . ' Description'
 457                                 && $this->mode[0] === self::MODE_STRUCT )
 458                  ) {
 459                         throw new MWException( "nesting mismatch. got a </$elm> but expected a </" . $this->curItem[0] . '>' );
 460                 }
 461
 462                 // Validate structures.
 463                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 464                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 465
 466                         $info =& $this->items[$ns][$tag];
 467                         $finalName = isset( $info['map_name'] )
 468                                 ? $info['map_name'] : $tag;
 469
 470                         $validate = is_array( $info['validate'] ) ? $info['validate']
 471                                 : array( 'XMPValidate', $info['validate'] );
 472
 473                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 474                                 // This can happen if all the members of the struct failed validation.
 475                                 wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> has no valid members." );
 476
 477                         } elseif ( is_callable( $validate ) ) {
 478                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 479                                 call_user_func_array( $validate, array( $info, &$val, false ) );
 480                                 if ( is_null( $val ) ) {
 481                                         // the idea being the validation function will unset the variable if
 482                                         // its invalid.
 483                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
 484                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 485                                 }
 486                         } else {
 487                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
 488                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 489                         }
 490                 }
 491
 492                 array_shift( $this->curItem );
 493                 array_shift( $this->mode );
 494                 $this->ancestorStruct = false;
 495                 $this->processingArray = false;
 496                 $this->itemLang = false;
 497         }
 498
 499         /**
 500         * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 501         * Add information about what type of element this is.
 502         *
 503         * Note we still have to hit the outer </property>
 504         *
 505         * For example, when processing:
 506         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 507         *   </rdf:Seq> </exif:ISOSpeedRatings>
 508         *
 509         * This method is called when we hit the </rdf:Seq>.
 510         * (For comparison, we call endElementModeSimple when we
 511         * hit the </rdf:li>)
 512         *
 513         * @param $elm String namespace . ' ' . element name
 514         */
 515         private function endElementModeLi( $elm ) {
 516
 517                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 518                 $info = $this->items[$ns][$tag];
 519                 $finalName = isset( $info['map_name'] )
 520                         ? $info['map_name'] : $tag;
 521
 522                 array_shift( $this->mode );
 523
 524                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 525                         wfDebugLog( 'XMP', __METHOD__ . " Empty compund element $finalName." );
 526                         return;
 527                 }
 528
 529                 if ( $elm === self::NS_RDF . ' Seq' ) {
 530                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 531                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 532                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 533                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 534                         // extra if needed as you could theoretically have a non-language alt.
 535                         if ( $info['mode'] === self::MODE_LANG ) {
 536                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 537                         }
 538
 539                 } else {
 540                         throw new MWException( __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
 541                 }
 542         }
 543         /**
 544         * End element while in MODE_QDESC
 545         * mostly when ending an element when we have a simple value
 546         * that has qualifiers.
 547         *
 548         * Qualifiers aren't all that common, and we don't do anything
 549         * with them.
 550         *
 551         * @param $elm String namespace and element
 552         */
 553         private function endElementModeQDesc( $elm ) {
 554
 555                 if ( $elm === self::NS_RDF . ' value' ) {
 556                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 557                         $this->saveValue( $ns, $tag, $this->charContent );
 558                         return;
 559                 } else {
 560                         array_shift( $this->mode );
 561                         array_shift( $this->curItem );
 562                 }
 563
 564
 565         }
 566         /**
 567         * Handler for hitting a closing element.
 568         *
 569         * generally just calls a helper function depending on what
 570         * mode we're in.
 571         *
 572         * Ignores the outer wrapping elements that are optional in
 573         * xmp and have no meaning.
 574         *
 575         * @param $parser XMLParser
 576         * @param $elm String namespace . ' ' . element name
 577         */
 578         function endElement( $parser, $elm ) {
 579                 if ( $elm === ( self::NS_RDF . ' RDF' )
 580                         || $elm === 'adobe:ns:meta/ xmpmeta'
 581                         || $elm === 'adobe:ns:meta/ xapmeta' )
 582                 {
 583                         // ignore these.
 584                         return;
 585                 }
 586
 587                 if ( $elm === self::NS_RDF . ' type' ) {
 588                         // these aren't really supported properly yet.
 589                         // However, it appears they almost never used.
 590                         wfDebugLog( 'XMP', __METHOD__ . ' encountered <rdf:type>' );
 591                 }
 592
 593                 if ( strpos( $elm, ' ' ) === false ) {
 594                         // This probably shouldn't happen.
 595                         // However, there is a bug in an adobe product
 596                         // that forgets the namespace on some things.
 597                         // (Luckily they are unimportant things).
 598                         wfDebugLog( 'XMP', __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 599                         return;
 600                 }
 601
 602                 if ( count( $this->mode[0] ) === 0 ) {
 603                         // This should never ever happen and means
 604                         // there is a pretty major bug in this class.
 605                         throw new MWException( 'Encountered end element with no mode' );
 606                 }
 607
 608                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 609                         // just to be paranoid. Should always have a curItem, except for initially
 610                         // (aka during MODE_INITAL).
 611                         throw new MWException( "Hit end element </$elm> but no curItem" );
 612                 }
 613
 614                 switch( $this->mode[0] ) {
 615                         case self::MODE_IGNORE:
 616                                 $this->endElementModeIgnore( $elm );
 617                                 break;
 618                         case self::MODE_SIMPLE:
 619                                 $this->endElementModeSimple( $elm );
 620                                 break;
 621                         case self::MODE_STRUCT:
 622                         case self::MODE_SEQ:
 623                         case self::MODE_BAG:
 624                         case self::MODE_LANG:
 625                         case self::MODE_BAGSTRUCT:
 626                                 $this->endElementNested( $elm );
 627                                 break;
 628                         case self::MODE_INITIAL:
 629                                 if ( $elm === self::NS_RDF . ' Description' ) {
 630                                         array_shift( $this->mode );
 631                                 } else {
 632                                         throw new MWException( 'Element ended unexpectedly while in MODE_INITIAL' );
 633                                 }
 634                                 break;
 635                         case self::MODE_LI:
 636                         case self::MODE_LI_LANG:
 637                                 $this->endElementModeLi( $elm );
 638                                 break;
 639                         case self::MODE_QDESC:
 640                                 $this->endElementModeQDesc( $elm );
 641                                 break;
 642                         default:
 643                                 wfDebugLog( 'XMP', __METHOD__ . " no mode (elm = $elm)" );
 644                                 break;
 645                 }
 646         }
 647
 648         /**
 649         * Hit an opening element while in MODE_IGNORE
 650         *
 651         * XMP is extensible, so ignore any tag we don't understand.
 652         *
 653         * Mostly ignores, unless we encounter the element that we are ignoring.
 654         * in which case we add it to the item stack, so we can ignore things
 655         * that are nested, correctly.
 656         *
 657         * @param $elm String namespace . ' ' . tag name
 658         */
 659         private function startElementModeIgnore( $elm ) {
 660                 if ( $elm === $this->curItem[0] ) {
 661                         array_unshift( $this->curItem, $elm );
 662                         array_unshift( $this->mode, self::MODE_IGNORE );
 663                 }
 664         }
 665
 666         /**
 667         *  Start element in MODE_BAG (unordered array)
 668         * this should always be <rdf:Bag>
 669         *
 670         * @param $elm String namespace . ' ' . tag
 671         * @throws MWException if we have an element that's not <rdf:Bag>
 672         */
 673         private function startElementModeBag( $elm ) {
 674                 if ( $elm === self::NS_RDF . ' Bag' ) {
 675                         array_unshift( $this->mode, self::MODE_LI );
 676                 } else {
 677                         throw new MWException( "Expected <rdf:Bag> but got $elm." );
 678                 }
 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         /**
 704         * Start element in MODE_LANG (language alternative)
 705         * this should always be <rdf:Alt>
 706         *
 707         * This tag tends to be used for metadata like describe this
 708         * picture, which can be translated into multiple languages.
 709         *
 710         * XMP supports non-linguistic alternative selections,
 711         * which are really only used for thumbnails, which
 712         * we don't care about.
 713         *
 714         * @param $elm String namespace . ' ' . tag
 715         * @throws MWException if we have an element that's not <rdf:Alt>
 716         */
 717         private function startElementModeLang( $elm ) {
 718                 if ( $elm === self::NS_RDF . ' Alt' ) {
 719                         array_unshift( $this->mode, self::MODE_LI_LANG );
 720                 } else {
 721                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 722                 }
 723
 724         }
 725
 726         /**
 727         * Handle an opening element when in MODE_SIMPLE
 728         *
 729         * This should not happen often. This is for if a simple element
 730         * already opened has a child element. Could happen for a
 731         * qualified element.
 732         *
 733         * For example:
 734         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 735         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 736         *   </exif:DigitalZoomRatio>
 737         *
 738         * This method is called when processing the <rdf:Description> element
 739         *
 740         * @param $elm String namespace and tag names separated by space.
 741         * @param $attribs Array Attributes of the element.
 742         */
 743         private function startElementModeSimple( $elm, $attribs ) {
 744                 if ( $elm === self::NS_RDF . ' Description' ) {
 745                         // If this value has qualifiers
 746                         array_unshift( $this->mode, self::MODE_QDESC );
 747                         array_unshift( $this->curItem, $this->curItem[0] );
 748
 749                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 750                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 751                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 752                         }
 753                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 754                         // This should not be here.
 755                         throw new MWException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 756
 757                 } else {
 758                         // something else we don't recognize, like a qualifier maybe.
 759                         wfDebugLog( 'XMP', __METHOD__ . " Encountered element <$elm> where only expecting character data as value of " . $this->curItem[0] );
 760                         array_unshift( $this->mode, self::MODE_IGNORE );
 761                         array_unshift( $this->curItem, $elm );
 762
 763                 }
 764
 765         }
 766
 767         /**
 768         * Start an element when in MODE_QDESC.
 769         * This generally happens when a simple element has an inner
 770         * rdf:Description to hold qualifier elements.
 771         *
 772         * For example in:
 773         * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 774         *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 775         *   </exif:DigitalZoomRatio>
 776         * Called when processing the <rdf:value> or <foo:someQualifier>.
 777         *
 778         * @param $elm String namespace and tag name separated by a space.
 779         *
 780         */
 781         private function startElementModeQDesc( $elm ) {
 782                 if ( $elm === self::NS_RDF . ' value' ) {
 783                         return; // do nothing
 784                 } else {
 785                         // otherwise its a qualifier, which we ignore
 786                         array_unshift( $this->mode, self::MODE_IGNORE );
 787                         array_unshift( $this->curItem, $elm );
 788                 }
 789         }
 790
 791         /**
 792         * Starting an element when in MODE_INITIAL
 793         * This usually happens when we hit an element inside
 794         * the outer rdf:Description
 795         *
 796         * This is generally where most properties start.
 797         *
 798         * @param $ns String Namespace
 799         * @param $tag String tag name (without namespace prefix)
 800         * @param $attribs Array array of attributes
 801         */
 802         private function startElementModeInitial( $ns, $tag, $attribs ) {
 803                 if ( $ns !== self::NS_RDF ) {
 804
 805                         if ( isset( $this->items[$ns][$tag] ) ) {
 806                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
 807                                         // If this element is supposed to appear only as
 808                                         // a child of a structure, but appears here (not as
 809                                         // a child of a struct), then something weird is
 810                                         // happening, so ignore this element and its children.
 811
 812                                         wfDebugLog( 'XMP', "Encountered <$ns:$tag> outside"
 813                                                 . " of its expected parent. Ignoring." );
 814
 815                                         array_unshift( $this->mode, self::MODE_IGNORE );
 816                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
 817                                         return;
 818                                 }
 819                                 $mode = $this->items[$ns][$tag]['mode'];
 820                                 array_unshift( $this->mode, $mode );
 821                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 822                                 if ( $mode === self::MODE_STRUCT ) {
 823                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
 824                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
 825                                 }
 826                                 if ( $this->charContent !== false ) {
 827                                         // Something weird.
 828                                         // Should not happen in valid XMP.
 829                                         throw new MWException( 'tag nested in non-whitespace characters.' );
 830                                 }
 831                         } else {
 832                                 // This element is not on our list of allowed elements so ignore.
 833                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
 834                                 array_unshift( $this->mode, self::MODE_IGNORE );
 835                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 836                                 return;
 837                         }
 838
 839                 }
 840                 // process attributes
 841                 $this->doAttribs( $attribs );
 842         }
 843
 844         /**
 845         * Hit an opening element when in a Struct (MODE_STRUCT)
 846         * This is generally for fields of a compound property.
 847         *
 848         * Example of a struct (abbreviated; flash has more properties):
 849         *
 850         * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
 851         *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
 852         *
 853         * or:
 854         *
 855         * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
 856         *  <exif:Mode>1</exif:Mode></exif:Flash>
 857         *
 858         * @param $ns String namespace
 859         * @param $tag String tag name (no ns)
 860         * @param $attribs Array array of attribs w/ values.
 861         */
 862         private function startElementModeStruct( $ns, $tag, $attribs ) {
 863                 if ( $ns !== self::NS_RDF ) {
 864
 865                         if ( isset( $this->items[$ns][$tag] ) ) {
 866                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
 867                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] ) )
 868                                 {
 869                                         // This assumes that we don't have inter-namespace nesting
 870                                         // which we don't in all the properties we're interested in.
 871                                         throw new MWException( " <$tag> appeared nested in <" . $this->ancestorStruct
 872                                                 . "> where it is not allowed." );
 873                                 }
 874                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
 875                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 876                                 if ( $this->charContent !== false ) {
 877                                         // Something weird.
 878                                         // Should not happen in valid XMP.
 879                                         throw new MWException( "tag <$tag> nested in non-whitespace characters (" . $this->charContent . ")." );
 880                                 }
 881                         } else {
 882                                 array_unshift( $this->mode, self::MODE_IGNORE );
 883                                 array_unshift( $this->curItem, $elm );
 884                                 return;
 885                         }
 886
 887                 }
 888
 889                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
 890                         $this->doAttribs( $attribs );
 891                         array_unshift( $this->mode, self::MODE_STRUCT );
 892                         array_unshift( $this->curItem, $this->curItem[0] );
 893                 }
 894         }
 895
 896         /**
 897         * opening element in MODE_LI
 898         * process elements of arrays.
 899         *
 900         * Example:
 901         * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 902         *   </rdf:Seq> </exif:ISOSpeedRatings>
 903         * This method is called when we hit the <rdf:li> element.
 904         *
 905         * @param $elm String: namespace . ' ' . tagname
 906         * @param $attribs Array: Attributes. (needed for BAGSTRUCTS)
 907         * @throws MWException if gets a tag other than <rdf:li>
 908         */
 909         private function startElementModeLi( $elm, $attribs ) {
 910                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
 911                         throw new MWException( "<rdf:li> expected but got $elm." );
 912                 }
 913
 914                 if ( !isset( $this->mode[1] ) ) {
 915                         // This should never ever ever happen. Checking for it
 916                         // to be paranoid.
 917                         throw new MWException( 'In mode Li, but no 2xPrevious mode!' );
 918                 }
 919
 920                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
 921                         // This list item contains a compound (STRUCT) value.
 922                         array_unshift( $this->mode, self::MODE_STRUCT );
 923                         array_unshift( $this->curItem, $elm );
 924                         $this->processingArray = true;
 925
 926                         if ( !isset( $this->curItem[1] ) ) {
 927                                 // be paranoid.
 928                                 throw new MWException( 'Can not find parent of BAGSTRUCT.' );
 929                         }
 930                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
 931                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
 932                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
 933
 934                         $this->doAttribs( $attribs );
 935
 936                 } else {
 937                         // Normal BAG or SEQ containing simple values.
 938                         array_unshift( $this->mode, self::MODE_SIMPLE );
 939                         // need to add curItem[0] on again since one is for the specific item
 940                         // and one is for the entire group.
 941                         array_unshift( $this->curItem, $this->curItem[0] );
 942                         $this->processingArray = true;
 943                 }
 944
 945         }
 946
 947         /**
 948         * Opening element in MODE_LI_LANG.
 949         * process elements of language alternatives
 950         *
 951         * Example:
 952         * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
 953         *  </rdf:li> </rdf:Alt> </dc:title>
 954         *
 955         * This method is called when we hit the <rdf:li> element.
 956         *
 957         * @param $elm String namespace . ' ' . tag
 958         * @param $attribs array array of elements (most importantly xml:lang)
 959         * @throws MWException if gets a tag other than <rdf:li> or if no xml:lang
 960         */
 961         private function startElementModeLiLang( $elm, $attribs ) {
 962                 if ( $elm !== self::NS_RDF . ' li' ) {
 963                         throw new MWException( __METHOD__ . " <rdf:li> expected but got $elm." );
 964                 }
 965                 if ( !isset( $attribs[ self::NS_XML . ' lang'] )
 966                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[ self::NS_XML . ' lang' ] ) )
 967                 {
 968                         throw new MWException( __METHOD__
 969                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
 970                 }
 971
 972                 // Lang is case-insensitive.
 973                 $this->itemLang = strtolower( $attribs[ self::NS_XML . ' lang' ] );
 974
 975                 // need to add curItem[0] on again since one is for the specific item
 976                 // and one is for the entire group.
 977                 array_unshift( $this->curItem, $this->curItem[0] );
 978                 array_unshift( $this->mode, self::MODE_SIMPLE );
 979                 $this->processingArray = true;
 980         }
 981
 982         /**
 983         * Hits an opening element.
 984         * Generally just calls a helper based on what MODE we're in.
 985         * Also does some initial set up for the wrapper element
 986         *
 987         * @param $parser XMLParser
 988         * @param $elm String namespace <space> element
 989         * @param $attribs Array attribute name => value
 990         */
 991         function startElement( $parser, $elm, $attribs ) {
 992
 993                 if ( $elm === self::NS_RDF . ' RDF'
 994                         || $elm === 'adobe:ns:meta/ xmpmeta'
 995                         || $elm === 'adobe:ns:meta/ xapmeta')
 996                 {
 997                         /* ignore. */
 998                         return;
 999                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1000                         if ( count( $this->mode ) === 0 ) {
1001                                 // outer rdf:desc
1002                                 array_unshift( $this->mode, self::MODE_INITIAL );
1003                         }
1004                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1005                         // This doesn't support rdf:type properly.
1006                         // In practise I have yet to see a file that
1007                         // uses this element, however it is mentioned
1008                         // on page 25 of part 1 of the xmp standard.
1009                         //
1010                         // also it seems as if exiv2 and exiftool do not support
1011                         // this either (That or I misunderstand the standard)
1012                         wfDebugLog( 'XMP', __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1013                 }
1014
1015                 if ( strpos( $elm, ' ' ) === false ) {
1016                         // This probably shouldn't happen.
1017                         wfDebugLog( 'XMP', __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1018                         return;
1019                 }
1020
1021                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1022
1023                 if ( count( $this->mode ) === 0 ) {
1024                         // This should not happen.
1025                         throw new MWException('Error extracting XMP, '
1026                                 . "encountered <$elm> with no mode" );
1027                 }
1028
1029                 switch( $this->mode[0] ) {
1030                         case self::MODE_IGNORE:
1031                                 $this->startElementModeIgnore( $elm );
1032                                 break;
1033                         case self::MODE_SIMPLE:
1034                                 $this->startElementModeSimple( $elm, $attribs );
1035                                 break;
1036                         case self::MODE_INITIAL:
1037                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1038                                 break;
1039                         case self::MODE_STRUCT:
1040                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1041                                 break;
1042                         case self::MODE_BAG:
1043                         case self::MODE_BAGSTRUCT:
1044                                 $this->startElementModeBag( $elm );
1045                                 break;
1046                         case self::MODE_SEQ:
1047                                 $this->startElementModeSeq( $elm );
1048                                 break;
1049                         case self::MODE_LANG:
1050                                 $this->startElementModeLang( $elm );
1051                                 break;
1052                         case self::MODE_LI_LANG:
1053                                 $this->startElementModeLiLang( $elm, $attribs );
1054                                 break;
1055                         case self::MODE_LI:
1056                                 $this->startElementModeLi( $elm, $attribs );
1057                                 break;
1058                         case self::MODE_QDESC:
1059                                 $this->startElementModeQDesc( $elm );
1060                                 break;
1061                         default:
1062                                 throw new MWException( 'StartElement in unknown mode: ' . $this->mode[0] );
1063                                 break;
1064                 }
1065         }
1066         /**
1067         * Process attributes.
1068         * Simple values can be stored as either a tag or attribute
1069         *
1070         * Often the initial <rdf:Description> tag just has all the simple
1071         * properties as attributes.
1072         *
1073         * Example:
1074         * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1075         *
1076         * @param $attribs Array attribute=>value array.
1077         */
1078         private function doAttribs( $attribs ) {
1079
1080                 // first check for rdf:parseType attribute, as that can change
1081                 // how the attributes are interperted.
1082
1083                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1084                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1085                         && $this->mode[0] === self::MODE_SIMPLE )
1086                 {
1087                         // this is equivalent to having an inner rdf:Description
1088                         $this->mode[0] = self::MODE_QDESC;
1089                 }
1090                 foreach ( $attribs as $name => $val ) {
1091
1092
1093                         if ( strpos( $name, ' ' ) === false ) {
1094                                 // This shouldn't happen, but so far some old software forgets namespace
1095                                 // on rdf:about.
1096                                 wfDebugLog( 'XMP', __METHOD__ . ' Encountered non-namespaced attribute: '
1097                                         . " $name=\"$val\". Skipping. " );
1098                                 continue;
1099                         }
1100                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1101                         if ( $ns === self::NS_RDF ) {
1102                                 if ( $tag === 'value' || $tag === 'resource' ) {
1103                                         // resource is for url.
1104                                         // value attribute is a weird way of just putting the contents.
1105                                         $this->char( $this->xmlParser, $val );
1106                                 }
1107                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1108                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1109                                         throw new MWException( __METHOD__
1110                                                 . " $ns:$tag found as attribute where not allowed" );
1111                                 }
1112                                 $this->saveValue( $ns, $tag, $val );
1113                         } else {
1114                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1115                         }
1116                 }
1117         }
1118         /**
1119         * Given an extracted value, save it to results array
1120         *
1121         * note also uses $this->ancestorStruct and
1122         * $this->processingArray to determine what name to
1123         * save the value under. (in addition to $tag).
1124         *
1125         * @param $ns String namespace of tag this is for
1126         * @param $tag String tag name
1127         * @param $val String value to save
1128         */
1129         private function saveValue( $ns, $tag, $val ) {
1130
1131                 $info =& $this->items[$ns][$tag];
1132                 $finalName = isset( $info['map_name'] )
1133                         ? $info['map_name'] : $tag;
1134                 if ( isset( $info['validate'] ) ) {
1135                         $validate = is_array( $info['validate'] ) ? $info['validate']
1136                                 : array( 'XMPValidate', $info['validate'] );
1137
1138                         if ( is_callable( $validate ) ) {
1139                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1140                                 // the reasoning behind using &$val instead of using the return value
1141                                 // is to be consistent between here and validating structures.
1142                                 if ( is_null( $val ) ) {
1143                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
1144                                         return;
1145                                 }
1146                         } else {
1147                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
1148                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1149                         }
1150                 }
1151
1152                 if ( $this->ancestorStruct && $this->processingArray ) {
1153                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1154                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1155                 } elseif ( $this->ancestorStruct ) {
1156                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1157                 } elseif ( $this->processingArray ) {
1158                         if ( $this->itemLang === false ) {
1159                                 // normal array
1160                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1161                         } else {
1162                                 // lang array.
1163                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1164                         }
1165                 } else {
1166                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1167                 }
1168         }
1169 }