includes/media/XMP.php

   1 <?php
   2 /**
   3  * Reader for XMP data containing properties relevant to images.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Media
  22  */
  23
  24 use Psr\Log\LoggerAwareInterface;
  25 use Psr\Log\LoggerInterface;
  26 use Psr\Log\NullLogger;
  27
  28 /**
  29  * Class for reading xmp data containing properties relevant to
  30  * images, and spitting out an array that FormatMetadata accepts.
  31  *
  32  * Note, this is not meant to recognize every possible thing you can
  33  * encode in XMP. It should recognize all the properties we want.
  34  * For example it doesn't have support for structures with multiple
  35  * nesting levels, as none of the properties we're supporting use that
  36  * feature. If it comes across properties it doesn't recognize, it should
  37  * ignore them.
  38  *
  39  * The public methods one would call in this class are
  40  * - parse( $content )
  41  *    Reads in xmp content.
  42  *    Can potentially be called multiple times with partial data each time.
  43  * - parseExtended( $content )
  44  *    Reads XMPExtended blocks (jpeg files only).
  45  * - getResults
  46  *    Outputs a results array.
  47  *
  48  * Note XMP kind of looks like rdf. They are not the same thing - XMP is
  49  * encoded as a specific subset of rdf. This class can read XMP. It cannot
  50  * read rdf.
  51  *
  52  */
  53 class XMPReader implements LoggerAwareInterface {
  54         /** @var array XMP item configuration array */
  55         protected $items;
  56
  57         /** @var array Array to hold the current element (and previous element, and so on) */
  58         private $curItem = array();
  59
  60         /** @var bool|string The structure name when processing nested structures. */
  61         private $ancestorStruct = false;
  62
  63         /** @var bool|string Temporary holder for character data that appears in xmp doc. */
  64         private $charContent = false;
  65
  66         /** @var array Stores the state the xmpreader is in (see MODE_FOO constants) */
  67         private $mode = array();
  68
  69         /** @var array Array to hold results */
  70         private $results = array();
  71
  72         /** @var bool If we're doing a seq or bag. */
  73         private $processingArray = false;
  74
  75         /** @var bool|string Used for lang alts only */
  76         private $itemLang = false;
  77
  78         /** @var resource A resource handle for the XML parser */
  79         private $xmlParser;
  80
  81         /** @var bool|string Character set like 'UTF-8' */
  82         private $charset = false;
  83
  84         /** @var int */
  85         private $extendedXMPOffset = 0;
  86
  87         /** @var int Flag determining if the XMP is safe to parse **/
  88         private $parsable = 0;
  89
  90         /** @var string Buffer of XML to parse **/
  91         private $xmlParsableBuffer = '';
  92
  93         /**
  94          * These are various mode constants.
  95          * they are used to figure out what to do
  96          * with an element when its encountered.
  97          *
  98          * For example, MODE_IGNORE is used when processing
  99          * a property we're not interested in. So if a new
 100          * element pops up when we're in that mode, we ignore it.
 101          */
 102         const MODE_INITIAL = 0;
 103         const MODE_IGNORE = 1;
 104         const MODE_LI = 2;
 105         const MODE_LI_LANG = 3;
 106         const MODE_QDESC = 4;
 107
 108         // The following MODE constants are also used in the
 109         // $items array to denote what type of property the item is.
 110         const MODE_SIMPLE = 10;
 111         const MODE_STRUCT = 11; // structure (associative array)
 112         const MODE_SEQ = 12; // ordered list
 113         const MODE_BAG = 13; // unordered list
 114         const MODE_LANG = 14;
 115         const MODE_ALT = 15; // non-language alt. Currently not implemented, and not needed atm.
 116         const MODE_BAGSTRUCT = 16; // A BAG of Structs.
 117
 118         const NS_RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
 119         const NS_XML = 'http://www.w3.org/XML/1998/namespace';
 120
 121         // States used while determining if XML is safe to parse
 122         const PARSABLE_UNKNOWN = 0;
 123         const PARSABLE_OK = 1;
 124         const PARSABLE_BUFFERING = 2;
 125         const PARSABLE_NO = 3;
 126
 127         /**
 128          * @var LoggerInterface
 129          */
 130         private $logger;
 131
 132         /**
 133          * Constructor.
 134          *
 135          * Primary job is to initialize the XMLParser
 136          */
 137         function __construct( LoggerInterface $logger = null ) {
 138
 139                 if ( !function_exists( 'xml_parser_create_ns' ) ) {
 140                         // this should already be checked by this point
 141                         throw new RuntimeException( 'XMP support requires XML Parser' );
 142                 }
 143                 if ( $logger ) {
 144                         $this->setLogger( $logger );
 145                 } else {
 146                         $this->setLogger( new NullLogger() );
 147                 }
 148
 149                 $this->items = XMPInfo::getItems();
 150
 151                 $this->resetXMLParser();
 152         }
 153
 154         public function setLogger( LoggerInterface $logger ) {
 155                 $this->logger = $logger;
 156         }
 157
 158         /**
 159          * free the XML parser.
 160          *
 161          * @note It is unclear to me if we really need to do this ourselves
 162          *  or if php garbage collection will automatically free the xmlParser
 163          *  when it is no longer needed.
 164          */
 165         private function destroyXMLParser() {
 166                 if ( $this->xmlParser ) {
 167                         xml_parser_free( $this->xmlParser );
 168                         $this->xmlParser = null;
 169                 }
 170         }
 171
 172         /**
 173          * Main use is if a single item has multiple xmp documents describing it.
 174          * For example in jpeg's with extendedXMP
 175          */
 176         private function resetXMLParser() {
 177
 178                 $this->destroyXMLParser();
 179
 180                 $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' );
 181                 xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 );
 182                 xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 );
 183
 184                 xml_set_element_handler( $this->xmlParser,
 185                         array( $this, 'startElement' ),
 186                         array( $this, 'endElement' ) );
 187
 188                 xml_set_character_data_handler( $this->xmlParser, array( $this, 'char' ) );
 189
 190                 $this->parsable = self::PARSABLE_UNKNOWN;
 191                 $this->xmlParsableBuffer = '';
 192         }
 193
 194         /**
 195          * Check if this instance supports using this class
 196          */
 197         public static function isSupported() {
 198                 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
 199         }
 200
 201         /** Get the result array. Do some post-processing before returning
 202          * the array, and transform any metadata that is special-cased.
 203          *
 204          * @return array Array of results as an array of arrays suitable for
 205          *    FormatMetadata::getFormattedData().
 206          */
 207         public function getResults() {
 208                 // xmp-special is for metadata that affects how stuff
 209                 // is extracted. For example xmpNote:HasExtendedXMP.
 210
 211                 // It is also used to handle photoshop:AuthorsPosition
 212                 // which is weird and really part of another property,
 213                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 214                 // The location fields also use it.
 215
 216                 $data = $this->results;
 217
 218                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 219                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 220                         && isset( $data['xmp-general']['Artist'][0] )
 221                 ) {
 222                         // Note, if there is more than one creator,
 223                         // this only applies to first. This also will
 224                         // only apply to the dc:Creator prop, not the
 225                         // exif:Artist prop.
 226
 227                         $data['xmp-general']['Artist'][0] =
 228                                 $data['xmp-special']['AuthorsPosition'] . ', '
 229                                 . $data['xmp-general']['Artist'][0];
 230                 }
 231
 232                 // Go through the LocationShown and LocationCreated
 233                 // changing it to the non-hierarchal form used by
 234                 // the other location fields.
 235
 236                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 237                         && is_array( $data['xmp-special']['LocationShown'][0] )
 238                 ) {
 239                         // the is_array is just paranoia. It should always
 240                         // be an array.
 241                         foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
 242                                 if ( !is_array( $loc ) ) {
 243                                         // To avoid copying over the _type meta-fields.
 244                                         continue;
 245                                 }
 246                                 foreach ( $loc as $field => $val ) {
 247                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 248                                 }
 249                         }
 250                 }
 251                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 252                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 253                 ) {
 254                         // the is_array is just paranoia. It should always
 255                         // be an array.
 256                         foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
 257                                 if ( !is_array( $loc ) ) {
 258                                         // To avoid copying over the _type meta-fields.
 259                                         continue;
 260                                 }
 261                                 foreach ( $loc as $field => $val ) {
 262                                         $data['xmp-general'][$field . 'Created'][] = $val;
 263                                 }
 264                         }
 265                 }
 266
 267                 // We don't want to return the special values, since they're
 268                 // special and not info to be stored about the file.
 269                 unset( $data['xmp-special'] );
 270
 271                 // Convert GPSAltitude to negative if below sea level.
 272                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
 273                         && isset( $data['xmp-exif']['GPSAltitude'] )
 274                 ) {
 275
 276                         // Must convert to a real before multiplying by -1
 277                         // XMPValidate guarantees there will always be a '/' in this value.
 278                         list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
 279                         $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
 280
 281                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
 282                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 283                         }
 284                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 285                 }
 286
 287                 return $data;
 288         }
 289
 290         /**
 291          * Main function to call to parse XMP. Use getResults to
 292          * get results.
 293          *
 294          * Also catches any errors during processing, writes them to
 295          * debug log, blanks result array and returns false.
 296          *
 297          * @param string $content XMP data
 298          * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
 299          * @throws RuntimeException
 300          * @return bool Success.
 301          */
 302         public function parse( $content, $allOfIt = true ) {
 303                 if ( !$this->xmlParser ) {
 304                         $this->resetXMLParser();
 305                 }
 306                 try {
 307
 308                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 309                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 310                         if ( !$this->charset ) {
 311                                 $bom = array();
 312                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 313                                         $content, $bom )
 314                                 ) {
 315                                         switch ( $bom[0] ) {
 316                                                 case "\xFE\xFF":
 317                                                         $this->charset = 'UTF-16BE';
 318                                                         break;
 319                                                 case "\xFF\xFE":
 320                                                         $this->charset = 'UTF-16LE';
 321                                                         break;
 322                                                 case "\x00\x00\xFE\xFF":
 323                                                         $this->charset = 'UTF-32BE';
 324                                                         break;
 325                                                 case "\xFF\xFE\x00\x00":
 326                                                         $this->charset = 'UTF-32LE';
 327                                                         break;
 328                                                 case "\xEF\xBB\xBF":
 329                                                         $this->charset = 'UTF-8';
 330                                                         break;
 331                                                 default:
 332                                                         //this should be impossible to get to
 333                                                         throw new RuntimeException( "Invalid BOM" );
 334                                         }
 335                                 } else {
 336                                         // standard specifically says, if no bom assume utf-8
 337                                         $this->charset = 'UTF-8';
 338                                 }
 339                         }
 340                         if ( $this->charset !== 'UTF-8' ) {
 341                                 //don't convert if already utf-8
 342                                 MediaWiki\suppressWarnings();
 343                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 344                                 MediaWiki\restoreWarnings();
 345                         }
 346
 347                         // Ensure the XMP block does not have an xml doctype declaration, which
 348                         // could declare entities unsafe to parse with xml_parse (T85848/T71210).
 349                         if ( $this->parsable !== self::PARSABLE_OK ) {
 350                                 if ( $this->parsable === self::PARSABLE_NO ) {
 351                                         throw new RuntimeException( 'Unsafe doctype declaration in XML.' );
 352                                 }
 353
 354                                 $content = $this->xmlParsableBuffer . $content;
 355                                 if ( !$this->checkParseSafety( $content ) ) {
 356                                         if ( !$allOfIt && $this->parsable !== self::PARSABLE_NO ) {
 357                                                 // parse wasn't Unsuccessful yet, so return true
 358                                                 // in this case.
 359                                                 return true;
 360                                         }
 361                                         $msg = ( $this->parsable === self::PARSABLE_NO ) ?
 362                                                 'Unsafe doctype declaration in XML.' :
 363                                                 'No root element found in XML.';
 364                                         throw new RuntimeException( $msg );
 365                                 }
 366                         }
 367
 368                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 369                         if ( !$ok ) {
 370                                 $code = xml_get_error_code( $this->xmlParser );
 371                                 $error = xml_error_string( $code );
 372                                 $line = xml_get_current_line_number( $this->xmlParser );
 373                                 $col = xml_get_current_column_number( $this->xmlParser );
 374                                 $offset = xml_get_current_byte_index( $this->xmlParser );
 375
 376                                 $this->logger->warning(
 377                                         '{method} : Error reading XMP content: {error} ' .
 378                                         '(line: {line} column: {column} byte offset: {offset})',
 379                                         array(
 380                                                 'method' => __METHOD__,
 381                                                 'error_code' => $code,
 382                                                 'error' => $error,
 383                                                 'line' => $line,
 384                                                 'column' => $col,
 385                                                 'offset' => $offset,
 386                                                 'content' => $content,
 387                                 ) );
 388                                 $this->results = array(); // blank if error.
 389                                 $this->destroyXMLParser();
 390                                 return false;
 391                         }
 392                 } catch ( Exception $e ) {
 393                         $this->logger->warning(
 394                                 '{method} Exception caught while parsing: ' . $e->getMessage(),
 395                                 array(
 396                                         'method' => __METHOD__,
 397                                         'exception' => $e,
 398                                         'content' => $content,
 399                                 )
 400                         );
 401                         $this->results = array();
 402                         return false;
 403                 }
 404                 if ( $allOfIt ) {
 405                         $this->destroyXMLParser();
 406                 }
 407
 408                 return true;
 409         }
 410
 411         /** Entry point for XMPExtended blocks in jpeg files
 412          *
 413          * @todo In serious need of testing
 414          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 415          * @param string $content XMPExtended block minus the namespace signature
 416          * @return bool If it succeeded.
 417          */
 418         public function parseExtended( $content ) {
 419                 // @todo FIXME: This is untested. Hard to find example files
 420                 // or programs that make such files..
 421                 $guid = substr( $content, 0, 32 );
 422                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 423                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid
 424                 ) {
 425                         $this->logger->info( __METHOD__ .
 426                                 " Ignoring XMPExtended block due to wrong guid (guid= '$guid')" );
 427
 428                         return false;
 429                 }
 430                 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 431
 432                 if ( !$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 433                         $this->logger->info( __METHOD__ . 'Error reading extended XMP block, invalid length or offset.' );
 434
 435                         return false;
 436                 }
 437
 438                 // we're not very robust here. we should accept it in the wrong order.
 439                 // To quote the XMP standard:
 440                 // "A JPEG writer should write the ExtendedXMP marker segments in order,
 441                 // immediately following the StandardXMP. However, the JPEG standard
 442                 // does not require preservation of marker segment order. A robust JPEG
 443                 // reader should tolerate the marker segments in any order."
 444                 //
 445                 // otoh the probability that an image will have more than 128k of
 446                 // metadata is rather low... so the probability that it will have
 447                 // > 128k, and be in the wrong order is very low...
 448
 449                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 450                         $this->logger->info( __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 451                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')' );
 452
 453                         return false;
 454                 }
 455
 456                 if ( $len['offset'] === 0 ) {
 457                         // if we're starting the extended block, we've probably already
 458                         // done the XMPStandard block, so reset.
 459                         $this->resetXMLParser();
 460                 }
 461
 462                 $this->extendedXMPOffset += $len['length'];
 463
 464                 $actualContent = substr( $content, 40 );
 465
 466                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 467                         $atEnd = true;
 468                 } else {
 469                         $atEnd = false;
 470                 }
 471
 472                 $this->logger->debug( __METHOD__ . 'Parsing a XMPExtended block' );
 473
 474                 return $this->parse( $actualContent, $atEnd );
 475         }
 476
 477         /**
 478          * Character data handler
 479          * Called whenever character data is found in the xmp document.
 480          *
 481          * does nothing if we're in MODE_IGNORE or if the data is whitespace
 482          * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 483          * data in the other modes).
 484          *
 485          * As an example, this happens when we encounter XMP like:
 486          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 487          * and are processing the 0/10 bit.
 488          *
 489          * @param XMLParser $parser XMLParser reference to the xml parser
 490          * @param string $data Character data
 491          * @throws RuntimeException On invalid data
 492          */
 493         function char( $parser, $data ) {
 494
 495                 $data = trim( $data );
 496                 if ( trim( $data ) === "" ) {
 497                         return;
 498                 }
 499
 500                 if ( !isset( $this->mode[0] ) ) {
 501                         throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
 502                 }
 503
 504                 if ( $this->mode[0] === self::MODE_IGNORE ) {
 505                         return;
 506                 }
 507
 508                 if ( $this->mode[0] !== self::MODE_SIMPLE
 509                         && $this->mode[0] !== self::MODE_QDESC
 510                 ) {
 511                         throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 512                 }
 513
 514                 // to check, how does this handle w.s.
 515                 if ( $this->charContent === false ) {
 516                         $this->charContent = $data;
 517                 } else {
 518                         $this->charContent .= $data;
 519                 }
 520         }
 521
 522         /**
 523          * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
 524          * contain a doctype declaration which could contain a dos attack if we
 525          * parse it and expand internal entities (T85848).
 526          *
 527          * @param string $content xml string to check for parse safety
 528          * @return bool true if the xml is safe to parse, false otherwise
 529          */
 530         private function checkParseSafety( $content ) {
 531                 $reader = new XMLReader();
 532                 $result = null;
 533
 534                 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
 535                 // instead of using XML().
 536                 $reader->open(
 537                         'data://text/plain,' . urlencode( $content ),
 538                         null,
 539                         LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
 540                 );
 541
 542                 $oldDisable = libxml_disable_entity_loader( true );
 543                 /** @noinspection PhpUnusedLocalVariableInspection */
 544                 $reset = new ScopedCallback(
 545                         'libxml_disable_entity_loader',
 546                         array( $oldDisable )
 547                 );
 548                 $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false );
 549
 550                 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
 551                 // when parsing truncated XML, which causes unit tests to fail.
 552                 MediaWiki\suppressWarnings();
 553                 while ( $reader->read() ) {
 554                         if ( $reader->nodeType === XMLReader::ELEMENT ) {
 555                                 // Reached the first element without hitting a doctype declaration
 556                                 $this->parsable = self::PARSABLE_OK;
 557                                 $result = true;
 558                                 break;
 559                         }
 560                         if ( $reader->nodeType === XMLReader::DOC_TYPE ) {
 561                                 $this->parsable = self::PARSABLE_NO;
 562                                 $result = false;
 563                                 break;
 564                         }
 565                 }
 566                 MediaWiki\restoreWarnings();
 567
 568                 if ( !is_null( $result ) ) {
 569                         return $result;
 570                 }
 571
 572                 // Reached the end of the parsable xml without finding an element
 573                 // or doctype. Buffer and try again.
 574                 $this->parsable = self::PARSABLE_BUFFERING;
 575                 $this->xmlParsableBuffer = $content;
 576                 return false;
 577         }
 578
 579         /** When we hit a closing element in MODE_IGNORE
 580          * Check to see if this is the element we started to ignore,
 581          * in which case we get out of MODE_IGNORE
 582          *
 583          * @param string $elm Namespace of element followed by a space and then tag name of element.
 584          */
 585         private function endElementModeIgnore( $elm ) {
 586                 if ( $this->curItem[0] === $elm ) {
 587                         array_shift( $this->curItem );
 588                         array_shift( $this->mode );
 589                 }
 590         }
 591
 592         /**
 593          * Hit a closing element when in MODE_SIMPLE.
 594          * This generally means that we finished processing a
 595          * property value, and now have to save the result to the
 596          * results array
 597          *
 598          * For example, when processing:
 599          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 600          * this deals with when we hit </exif:DigitalZoomRatio>.
 601          *
 602          * Or it could be if we hit the end element of a property
 603          * of a compound data structure (like a member of an array).
 604          *
 605          * @param string $elm Namespace, space, and tag name.
 606          */
 607         private function endElementModeSimple( $elm ) {
 608                 if ( $this->charContent !== false ) {
 609                         if ( $this->processingArray ) {
 610                                 // if we're processing an array, use the original element
 611                                 // name instead of rdf:li.
 612                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 613                         } else {
 614                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 615                         }
 616                         $this->saveValue( $ns, $tag, $this->charContent );
 617
 618                         $this->charContent = false; // reset
 619                 }
 620                 array_shift( $this->curItem );
 621                 array_shift( $this->mode );
 622         }
 623
 624         /**
 625          * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 626          * generally means we've finished processing a nested structure.
 627          * resets some internal variables to indicate that.
 628          *
 629          * Note this means we hit the closing element not the "</rdf:Seq>".
 630          *
 631          * @par For example, when processing:
 632          * @code{,xml}
 633          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 634          *   </rdf:Seq> </exif:ISOSpeedRatings>
 635          * @endcode
 636          *
 637          * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
 638          *
 639          * @param string $elm Namespace . space . tag name.
 640          * @throws RuntimeException
 641          */
 642         private function endElementNested( $elm ) {
 643
 644                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 645                    in which case it could also be rdf:Description */
 646                 if ( $this->curItem[0] !== $elm
 647                         && !( $elm === self::NS_RDF . ' Description'
 648                                 && $this->mode[0] === self::MODE_STRUCT )
 649                 ) {
 650                         throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
 651                                 $this->curItem[0] . '>' );
 652                 }
 653
 654                 // Validate structures.
 655                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 656                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 657                         $info =& $this->items[$ns][$tag];
 658                         $finalName = isset( $info['map_name'] )
 659                                 ? $info['map_name'] : $tag;
 660
 661                         if ( is_array( $info['validate'] ) ) {
 662                                 $validate = $info['validate'];
 663                         } else {
 664                                 $validator = new XMPValidate( $this->logger );
 665                                 $validate = array( $validator, $info['validate'] );
 666                         }
 667
 668                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 669                                 // This can happen if all the members of the struct failed validation.
 670                                 $this->logger->debug( __METHOD__ . " <$ns:$tag> has no valid members." );
 671                         } elseif ( is_callable( $validate ) ) {
 672                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 673                                 call_user_func_array( $validate, array( $info, &$val, false ) );
 674                                 if ( is_null( $val ) ) {
 675                                         // the idea being the validation function will unset the variable if
 676                                         // its invalid.
 677                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
 678                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 679                                 }
 680                         } else {
 681                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
 682                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 683                         }
 684                 }
 685
 686                 array_shift( $this->curItem );
 687                 array_shift( $this->mode );
 688                 $this->ancestorStruct = false;
 689                 $this->processingArray = false;
 690                 $this->itemLang = false;
 691         }
 692
 693         /**
 694          * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 695          * Add information about what type of element this is.
 696          *
 697          * Note we still have to hit the outer "</property>"
 698          *
 699          * @par For example, when processing:
 700          * @code{,xml}
 701          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 702          *   </rdf:Seq> </exif:ISOSpeedRatings>
 703          * @endcode
 704          *
 705          * This method is called when we hit the "</rdf:Seq>".
 706          * (For comparison, we call endElementModeSimple when we
 707          * hit the "</rdf:li>")
 708          *
 709          * @param string $elm Namespace . ' ' . element name
 710          * @throws RuntimeException
 711          */
 712         private function endElementModeLi( $elm ) {
 713
 714                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 715                 $info = $this->items[$ns][$tag];
 716                 $finalName = isset( $info['map_name'] )
 717                         ? $info['map_name'] : $tag;
 718
 719                 array_shift( $this->mode );
 720
 721                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 722                         $this->logger->debug( __METHOD__ . " Empty compund element $finalName." );
 723
 724                         return;
 725                 }
 726
 727                 if ( $elm === self::NS_RDF . ' Seq' ) {
 728                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 729                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 730                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 731                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 732                         // extra if needed as you could theoretically have a non-language alt.
 733                         if ( $info['mode'] === self::MODE_LANG ) {
 734                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 735                         }
 736                 } else {
 737                         throw new RuntimeException( __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
 738                 }
 739         }
 740
 741         /**
 742          * End element while in MODE_QDESC
 743          * mostly when ending an element when we have a simple value
 744          * that has qualifiers.
 745          *
 746          * Qualifiers aren't all that common, and we don't do anything
 747          * with them.
 748          *
 749          * @param string $elm Namespace and element
 750          */
 751         private function endElementModeQDesc( $elm ) {
 752
 753                 if ( $elm === self::NS_RDF . ' value' ) {
 754                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 755                         $this->saveValue( $ns, $tag, $this->charContent );
 756
 757                         return;
 758                 } else {
 759                         array_shift( $this->mode );
 760                         array_shift( $this->curItem );
 761                 }
 762         }
 763
 764         /**
 765          * Handler for hitting a closing element.
 766          *
 767          * generally just calls a helper function depending on what
 768          * mode we're in.
 769          *
 770          * Ignores the outer wrapping elements that are optional in
 771          * xmp and have no meaning.
 772          *
 773          * @param XMLParser $parser
 774          * @param string $elm Namespace . ' ' . element name
 775          * @throws RuntimeException
 776          */
 777         function endElement( $parser, $elm ) {
 778                 if ( $elm === ( self::NS_RDF . ' RDF' )
 779                         || $elm === 'adobe:ns:meta/ xmpmeta'
 780                         || $elm === 'adobe:ns:meta/ xapmeta'
 781                 ) {
 782                         // ignore these.
 783                         return;
 784                 }
 785
 786                 if ( $elm === self::NS_RDF . ' type' ) {
 787                         // these aren't really supported properly yet.
 788                         // However, it appears they almost never used.
 789                         $this->logger->info( __METHOD__ . ' encountered <rdf:type>' );
 790                 }
 791
 792                 if ( strpos( $elm, ' ' ) === false ) {
 793                         // This probably shouldn't happen.
 794                         // However, there is a bug in an adobe product
 795                         // that forgets the namespace on some things.
 796                         // (Luckily they are unimportant things).
 797                         $this->logger->info( __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 798
 799                         return;
 800                 }
 801
 802                 if ( count( $this->mode[0] ) === 0 ) {
 803                         // This should never ever happen and means
 804                         // there is a pretty major bug in this class.
 805                         throw new RuntimeException( 'Encountered end element with no mode' );
 806                 }
 807
 808                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 809                         // just to be paranoid. Should always have a curItem, except for initially
 810                         // (aka during MODE_INITAL).
 811                         throw new RuntimeException( "Hit end element </$elm> but no curItem" );
 812                 }
 813
 814                 switch ( $this->mode[0] ) {
 815                         case self::MODE_IGNORE:
 816                                 $this->endElementModeIgnore( $elm );
 817                                 break;
 818                         case self::MODE_SIMPLE:
 819                                 $this->endElementModeSimple( $elm );
 820                                 break;
 821                         case self::MODE_STRUCT:
 822                         case self::MODE_SEQ:
 823                         case self::MODE_BAG:
 824                         case self::MODE_LANG:
 825                         case self::MODE_BAGSTRUCT:
 826                                 $this->endElementNested( $elm );
 827                                 break;
 828                         case self::MODE_INITIAL:
 829                                 if ( $elm === self::NS_RDF . ' Description' ) {
 830                                         array_shift( $this->mode );
 831                                 } else {
 832                                         throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
 833                                 }
 834                                 break;
 835                         case self::MODE_LI:
 836                         case self::MODE_LI_LANG:
 837                                 $this->endElementModeLi( $elm );
 838                                 break;
 839                         case self::MODE_QDESC:
 840                                 $this->endElementModeQDesc( $elm );
 841                                 break;
 842                         default:
 843                                 $this->logger->warning( __METHOD__ . " no mode (elm = $elm)" );
 844                                 break;
 845                 }
 846         }
 847
 848         /**
 849          * Hit an opening element while in MODE_IGNORE
 850          *
 851          * XMP is extensible, so ignore any tag we don't understand.
 852          *
 853          * Mostly ignores, unless we encounter the element that we are ignoring.
 854          * in which case we add it to the item stack, so we can ignore things
 855          * that are nested, correctly.
 856          *
 857          * @param string $elm Namespace . ' ' . tag name
 858          */
 859         private function startElementModeIgnore( $elm ) {
 860                 if ( $elm === $this->curItem[0] ) {
 861                         array_unshift( $this->curItem, $elm );
 862                         array_unshift( $this->mode, self::MODE_IGNORE );
 863                 }
 864         }
 865
 866         /**
 867          *  Start element in MODE_BAG (unordered array)
 868          * this should always be <rdf:Bag>
 869          *
 870          * @param string $elm Namespace . ' ' . tag
 871          * @throws RuntimeException If we have an element that's not <rdf:Bag>
 872          */
 873         private function startElementModeBag( $elm ) {
 874                 if ( $elm === self::NS_RDF . ' Bag' ) {
 875                         array_unshift( $this->mode, self::MODE_LI );
 876                 } else {
 877                         throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
 878                 }
 879         }
 880
 881         /**
 882          * Start element in MODE_SEQ (ordered array)
 883          * this should always be <rdf:Seq>
 884          *
 885          * @param string $elm Namespace . ' ' . tag
 886          * @throws RuntimeException If we have an element that's not <rdf:Seq>
 887          */
 888         private function startElementModeSeq( $elm ) {
 889                 if ( $elm === self::NS_RDF . ' Seq' ) {
 890                         array_unshift( $this->mode, self::MODE_LI );
 891                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 892                         # bug 27105
 893                         $this->logger->info( __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 894                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 895                         array_unshift( $this->mode, self::MODE_LI );
 896                 } else {
 897                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 898                 }
 899         }
 900
 901         /**
 902          * Start element in MODE_LANG (language alternative)
 903          * this should always be <rdf:Alt>
 904          *
 905          * This tag tends to be used for metadata like describe this
 906          * picture, which can be translated into multiple languages.
 907          *
 908          * XMP supports non-linguistic alternative selections,
 909          * which are really only used for thumbnails, which
 910          * we don't care about.
 911          *
 912          * @param string $elm Namespace . ' ' . tag
 913          * @throws RuntimeException If we have an element that's not <rdf:Alt>
 914          */
 915         private function startElementModeLang( $elm ) {
 916                 if ( $elm === self::NS_RDF . ' Alt' ) {
 917                         array_unshift( $this->mode, self::MODE_LI_LANG );
 918                 } else {
 919                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 920                 }
 921         }
 922
 923         /**
 924          * Handle an opening element when in MODE_SIMPLE
 925          *
 926          * This should not happen often. This is for if a simple element
 927          * already opened has a child element. Could happen for a
 928          * qualified element.
 929          *
 930          * For example:
 931          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 932          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 933          *   </exif:DigitalZoomRatio>
 934          *
 935          * This method is called when processing the <rdf:Description> element
 936          *
 937          * @param string $elm Namespace and tag names separated by space.
 938          * @param array $attribs Attributes of the element.
 939          * @throws RuntimeException
 940          */
 941         private function startElementModeSimple( $elm, $attribs ) {
 942                 if ( $elm === self::NS_RDF . ' Description' ) {
 943                         // If this value has qualifiers
 944                         array_unshift( $this->mode, self::MODE_QDESC );
 945                         array_unshift( $this->curItem, $this->curItem[0] );
 946
 947                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 948                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 949                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 950                         }
 951                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 952                         // This should not be here.
 953                         throw new RuntimeException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 954                 } else {
 955                         // something else we don't recognize, like a qualifier maybe.
 956                         $this->logger->info( __METHOD__ .
 957                                 " Encountered element <$elm> where only expecting character data as value of " .
 958                                 $this->curItem[0] );
 959                         array_unshift( $this->mode, self::MODE_IGNORE );
 960                         array_unshift( $this->curItem, $elm );
 961                 }
 962         }
 963
 964         /**
 965          * Start an element when in MODE_QDESC.
 966          * This generally happens when a simple element has an inner
 967          * rdf:Description to hold qualifier elements.
 968          *
 969          * For example in:
 970          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 971          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 972          *   </exif:DigitalZoomRatio>
 973          * Called when processing the <rdf:value> or <foo:someQualifier>.
 974          *
 975          * @param string $elm Namespace and tag name separated by a space.
 976          *
 977          */
 978         private function startElementModeQDesc( $elm ) {
 979                 if ( $elm === self::NS_RDF . ' value' ) {
 980                         return; // do nothing
 981                 } else {
 982                         // otherwise its a qualifier, which we ignore
 983                         array_unshift( $this->mode, self::MODE_IGNORE );
 984                         array_unshift( $this->curItem, $elm );
 985                 }
 986         }
 987
 988         /**
 989          * Starting an element when in MODE_INITIAL
 990          * This usually happens when we hit an element inside
 991          * the outer rdf:Description
 992          *
 993          * This is generally where most properties start.
 994          *
 995          * @param string $ns Namespace
 996          * @param string $tag Tag name (without namespace prefix)
 997          * @param array $attribs Array of attributes
 998          * @throws RuntimeException
 999          */
1000         private function startElementModeInitial( $ns, $tag, $attribs ) {
1001                 if ( $ns !== self::NS_RDF ) {
1002
1003                         if ( isset( $this->items[$ns][$tag] ) ) {
1004                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
1005                                         // If this element is supposed to appear only as
1006                                         // a child of a structure, but appears here (not as
1007                                         // a child of a struct), then something weird is
1008                                         // happening, so ignore this element and its children.
1009
1010                                         $this->logger->warning( "Encountered <$ns:$tag> outside"
1011                                                 . " of its expected parent. Ignoring." );
1012
1013                                         array_unshift( $this->mode, self::MODE_IGNORE );
1014                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
1015
1016                                         return;
1017                                 }
1018                                 $mode = $this->items[$ns][$tag]['mode'];
1019                                 array_unshift( $this->mode, $mode );
1020                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1021                                 if ( $mode === self::MODE_STRUCT ) {
1022                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
1023                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
1024                                 }
1025                                 if ( $this->charContent !== false ) {
1026                                         // Something weird.
1027                                         // Should not happen in valid XMP.
1028                                         throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1029                                 }
1030                         } else {
1031                                 // This element is not on our list of allowed elements so ignore.
1032                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1033                                 array_unshift( $this->mode, self::MODE_IGNORE );
1034                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1035
1036                                 return;
1037                         }
1038                 }
1039                 // process attributes
1040                 $this->doAttribs( $attribs );
1041         }
1042
1043         /**
1044          * Hit an opening element when in a Struct (MODE_STRUCT)
1045          * This is generally for fields of a compound property.
1046          *
1047          * Example of a struct (abbreviated; flash has more properties):
1048          *
1049          * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1050          *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1051          *
1052          * or:
1053          *
1054          * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1055          *  <exif:Mode>1</exif:Mode></exif:Flash>
1056          *
1057          * @param string $ns Namespace
1058          * @param string $tag Tag name (no ns)
1059          * @param array $attribs Array of attribs w/ values.
1060          * @throws RuntimeException
1061          */
1062         private function startElementModeStruct( $ns, $tag, $attribs ) {
1063                 if ( $ns !== self::NS_RDF ) {
1064
1065                         if ( isset( $this->items[$ns][$tag] ) ) {
1066                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
1067                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] )
1068                                 ) {
1069                                         // This assumes that we don't have inter-namespace nesting
1070                                         // which we don't in all the properties we're interested in.
1071                                         throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1072                                                 . "> where it is not allowed." );
1073                                 }
1074                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
1075                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1076                                 if ( $this->charContent !== false ) {
1077                                         // Something weird.
1078                                         // Should not happen in valid XMP.
1079                                         throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1080                                                 $this->charContent . ")." );
1081                                 }
1082                         } else {
1083                                 array_unshift( $this->mode, self::MODE_IGNORE );
1084                                 array_unshift( $this->curItem, $elm );
1085
1086                                 return;
1087                         }
1088                 }
1089
1090                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
1091                         $this->doAttribs( $attribs );
1092                         array_unshift( $this->mode, self::MODE_STRUCT );
1093                         array_unshift( $this->curItem, $this->curItem[0] );
1094                 }
1095         }
1096
1097         /**
1098          * opening element in MODE_LI
1099          * process elements of arrays.
1100          *
1101          * Example:
1102          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1103          *   </rdf:Seq> </exif:ISOSpeedRatings>
1104          * This method is called when we hit the <rdf:li> element.
1105          *
1106          * @param string $elm Namespace . ' ' . tagname
1107          * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1108          * @throws RuntimeException If gets a tag other than <rdf:li>
1109          */
1110         private function startElementModeLi( $elm, $attribs ) {
1111                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
1112                         throw new RuntimeException( "<rdf:li> expected but got $elm." );
1113                 }
1114
1115                 if ( !isset( $this->mode[1] ) ) {
1116                         // This should never ever ever happen. Checking for it
1117                         // to be paranoid.
1118                         throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1119                 }
1120
1121                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
1122                         // This list item contains a compound (STRUCT) value.
1123                         array_unshift( $this->mode, self::MODE_STRUCT );
1124                         array_unshift( $this->curItem, $elm );
1125                         $this->processingArray = true;
1126
1127                         if ( !isset( $this->curItem[1] ) ) {
1128                                 // be paranoid.
1129                                 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1130                         }
1131                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
1132                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
1133                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
1134
1135                         $this->doAttribs( $attribs );
1136                 } else {
1137                         // Normal BAG or SEQ containing simple values.
1138                         array_unshift( $this->mode, self::MODE_SIMPLE );
1139                         // need to add curItem[0] on again since one is for the specific item
1140                         // and one is for the entire group.
1141                         array_unshift( $this->curItem, $this->curItem[0] );
1142                         $this->processingArray = true;
1143                 }
1144         }
1145
1146         /**
1147          * Opening element in MODE_LI_LANG.
1148          * process elements of language alternatives
1149          *
1150          * Example:
1151          * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1152          *  </rdf:li> </rdf:Alt> </dc:title>
1153          *
1154          * This method is called when we hit the <rdf:li> element.
1155          *
1156          * @param string $elm Namespace . ' ' . tag
1157          * @param array $attribs Array of elements (most importantly xml:lang)
1158          * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1159          */
1160         private function startElementModeLiLang( $elm, $attribs ) {
1161                 if ( $elm !== self::NS_RDF . ' li' ) {
1162                         throw new RuntimeException( __METHOD__ . " <rdf:li> expected but got $elm." );
1163                 }
1164                 if ( !isset( $attribs[self::NS_XML . ' lang'] )
1165                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] )
1166                 ) {
1167                         throw new RuntimeException( __METHOD__
1168                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1169                 }
1170
1171                 // Lang is case-insensitive.
1172                 $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] );
1173
1174                 // need to add curItem[0] on again since one is for the specific item
1175                 // and one is for the entire group.
1176                 array_unshift( $this->curItem, $this->curItem[0] );
1177                 array_unshift( $this->mode, self::MODE_SIMPLE );
1178                 $this->processingArray = true;
1179         }
1180
1181         /**
1182          * Hits an opening element.
1183          * Generally just calls a helper based on what MODE we're in.
1184          * Also does some initial set up for the wrapper element
1185          *
1186          * @param XMLParser $parser
1187          * @param string $elm Namespace "<space>" element
1188          * @param array $attribs Attribute name => value
1189          * @throws RuntimeException
1190          */
1191         function startElement( $parser, $elm, $attribs ) {
1192
1193                 if ( $elm === self::NS_RDF . ' RDF'
1194                         || $elm === 'adobe:ns:meta/ xmpmeta'
1195                         || $elm === 'adobe:ns:meta/ xapmeta'
1196                 ) {
1197                         /* ignore. */
1198                         return;
1199                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1200                         if ( count( $this->mode ) === 0 ) {
1201                                 // outer rdf:desc
1202                                 array_unshift( $this->mode, self::MODE_INITIAL );
1203                         }
1204                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1205                         // This doesn't support rdf:type properly.
1206                         // In practise I have yet to see a file that
1207                         // uses this element, however it is mentioned
1208                         // on page 25 of part 1 of the xmp standard.
1209                         //
1210                         // also it seems as if exiv2 and exiftool do not support
1211                         // this either (That or I misunderstand the standard)
1212                         $this->logger->info( __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1213                 }
1214
1215                 if ( strpos( $elm, ' ' ) === false ) {
1216                         // This probably shouldn't happen.
1217                         $this->logger->info( __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1218
1219                         return;
1220                 }
1221
1222                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1223
1224                 if ( count( $this->mode ) === 0 ) {
1225                         // This should not happen.
1226                         throw new RuntimeException( 'Error extracting XMP, '
1227                                 . "encountered <$elm> with no mode" );
1228                 }
1229
1230                 switch ( $this->mode[0] ) {
1231                         case self::MODE_IGNORE:
1232                                 $this->startElementModeIgnore( $elm );
1233                                 break;
1234                         case self::MODE_SIMPLE:
1235                                 $this->startElementModeSimple( $elm, $attribs );
1236                                 break;
1237                         case self::MODE_INITIAL:
1238                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1239                                 break;
1240                         case self::MODE_STRUCT:
1241                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1242                                 break;
1243                         case self::MODE_BAG:
1244                         case self::MODE_BAGSTRUCT:
1245                                 $this->startElementModeBag( $elm );
1246                                 break;
1247                         case self::MODE_SEQ:
1248                                 $this->startElementModeSeq( $elm );
1249                                 break;
1250                         case self::MODE_LANG:
1251                                 $this->startElementModeLang( $elm );
1252                                 break;
1253                         case self::MODE_LI_LANG:
1254                                 $this->startElementModeLiLang( $elm, $attribs );
1255                                 break;
1256                         case self::MODE_LI:
1257                                 $this->startElementModeLi( $elm, $attribs );
1258                                 break;
1259                         case self::MODE_QDESC:
1260                                 $this->startElementModeQDesc( $elm );
1261                                 break;
1262                         default:
1263                                 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode[0] );
1264                 }
1265         }
1266
1267         /**
1268          * Process attributes.
1269          * Simple values can be stored as either a tag or attribute
1270          *
1271          * Often the initial "<rdf:Description>" tag just has all the simple
1272          * properties as attributes.
1273          *
1274          * @codingStandardsIgnoreStart Long line that cannot be broken
1275          * @par Example:
1276          * @code
1277          * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1278          * @endcode
1279          * @codingStandardsIgnoreEnd
1280          *
1281          * @param array $attribs Array attribute=>value
1282          * @throws RuntimeException
1283          */
1284         private function doAttribs( $attribs ) {
1285                 // first check for rdf:parseType attribute, as that can change
1286                 // how the attributes are interperted.
1287
1288                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1289                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1290                         && $this->mode[0] === self::MODE_SIMPLE
1291                 ) {
1292                         // this is equivalent to having an inner rdf:Description
1293                         $this->mode[0] = self::MODE_QDESC;
1294                 }
1295                 foreach ( $attribs as $name => $val ) {
1296                         if ( strpos( $name, ' ' ) === false ) {
1297                                 // This shouldn't happen, but so far some old software forgets namespace
1298                                 // on rdf:about.
1299                                 $this->logger->info( __METHOD__ . ' Encountered non-namespaced attribute: '
1300                                         . " $name=\"$val\". Skipping. " );
1301                                 continue;
1302                         }
1303                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1304                         if ( $ns === self::NS_RDF ) {
1305                                 if ( $tag === 'value' || $tag === 'resource' ) {
1306                                         // resource is for url.
1307                                         // value attribute is a weird way of just putting the contents.
1308                                         $this->char( $this->xmlParser, $val );
1309                                 }
1310                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1311                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1312                                         throw new RuntimeException( __METHOD__
1313                                                 . " $ns:$tag found as attribute where not allowed" );
1314                                 }
1315                                 $this->saveValue( $ns, $tag, $val );
1316                         } else {
1317                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1318                         }
1319                 }
1320         }
1321
1322         /**
1323          * Given an extracted value, save it to results array
1324          *
1325          * note also uses $this->ancestorStruct and
1326          * $this->processingArray to determine what name to
1327          * save the value under. (in addition to $tag).
1328          *
1329          * @param string $ns Namespace of tag this is for
1330          * @param string $tag Tag name
1331          * @param string $val Value to save
1332          */
1333         private function saveValue( $ns, $tag, $val ) {
1334
1335                 $info =& $this->items[$ns][$tag];
1336                 $finalName = isset( $info['map_name'] )
1337                         ? $info['map_name'] : $tag;
1338                 if ( isset( $info['validate'] ) ) {
1339                         if ( is_array( $info['validate'] ) ) {
1340                                 $validate = $info['validate'];
1341                         } else {
1342                                 $validator = new XMPValidate( $this->logger );
1343                                 $validate = array( $validator, $info['validate'] );
1344                         }
1345
1346                         if ( is_callable( $validate ) ) {
1347                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1348                                 // the reasoning behind using &$val instead of using the return value
1349                                 // is to be consistent between here and validating structures.
1350                                 if ( is_null( $val ) ) {
1351                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
1352
1353                                         return;
1354                                 }
1355                         } else {
1356                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
1357                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1358                         }
1359                 }
1360
1361                 if ( $this->ancestorStruct && $this->processingArray ) {
1362                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1363                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1364                 } elseif ( $this->ancestorStruct ) {
1365                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1366                 } elseif ( $this->processingArray ) {
1367                         if ( $this->itemLang === false ) {
1368                                 // normal array
1369                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1370                         } else {
1371                                 // lang array.
1372                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1373                         }
1374                 } else {
1375                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1376                 }
1377         }
1378 }