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 Exception( '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 Exception( $msg );
 365                                 }
 366                         }
 367
 368                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 369                         if ( !$ok ) {
 370                                 $error = xml_error_string( xml_get_error_code( $this->xmlParser ) );
 371                                 $where = 'line: ' . xml_get_current_line_number( $this->xmlParser )
 372                                         . ' column: ' . xml_get_current_column_number( $this->xmlParser )
 373                                         . ' byte offset: ' . xml_get_current_byte_index( $this->xmlParser );
 374
 375                                 $this->logger->info( "XMPReader::parse : Error reading XMP content: $error ($where)" );
 376                                 $this->results = array(); // blank if error.
 377                                 $this->destroyXMLParser();
 378                                 return false;
 379                         }
 380                 } catch ( Exception $e ) {
 381                         $this->logger->info( 'XMP parse error: ' . $e );
 382                         $this->results = array();
 383
 384                         if ( $allOfIt ) {
 385                                 $this->destroyXMLParser();
 386                         }
 387                         return false;
 388                 }
 389                 if ( $allOfIt ) {
 390                         $this->destroyXMLParser();
 391                 }
 392
 393                 return true;
 394         }
 395
 396         /** Entry point for XMPExtended blocks in jpeg files
 397          *
 398          * @todo In serious need of testing
 399          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 400          * @param string $content XMPExtended block minus the namespace signature
 401          * @return bool If it succeeded.
 402          */
 403         public function parseExtended( $content ) {
 404                 // @todo FIXME: This is untested. Hard to find example files
 405                 // or programs that make such files..
 406                 $guid = substr( $content, 0, 32 );
 407                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 408                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid
 409                 ) {
 410                         $this->logger->info( __METHOD__ .
 411                                 " Ignoring XMPExtended block due to wrong guid (guid= '$guid')" );
 412
 413                         return false;
 414                 }
 415                 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 416
 417                 if ( !$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 418                         $this->logger->info( __METHOD__ . 'Error reading extended XMP block, invalid length or offset.' );
 419
 420                         return false;
 421                 }
 422
 423                 // we're not very robust here. we should accept it in the wrong order.
 424                 // To quote the XMP standard:
 425                 // "A JPEG writer should write the ExtendedXMP marker segments in order,
 426                 // immediately following the StandardXMP. However, the JPEG standard
 427                 // does not require preservation of marker segment order. A robust JPEG
 428                 // reader should tolerate the marker segments in any order."
 429                 //
 430                 // otoh the probability that an image will have more than 128k of
 431                 // metadata is rather low... so the probability that it will have
 432                 // > 128k, and be in the wrong order is very low...
 433
 434                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 435                         $this->logger->info( __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 436                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')' );
 437
 438                         return false;
 439                 }
 440
 441                 if ( $len['offset'] === 0 ) {
 442                         // if we're starting the extended block, we've probably already
 443                         // done the XMPStandard block, so reset.
 444                         $this->resetXMLParser();
 445                 }
 446
 447                 $this->extendedXMPOffset += $len['length'];
 448
 449                 $actualContent = substr( $content, 40 );
 450
 451                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 452                         $atEnd = true;
 453                 } else {
 454                         $atEnd = false;
 455                 }
 456
 457                 $this->logger->debug( __METHOD__ . 'Parsing a XMPExtended block' );
 458
 459                 return $this->parse( $actualContent, $atEnd );
 460         }
 461
 462         /**
 463          * Character data handler
 464          * Called whenever character data is found in the xmp document.
 465          *
 466          * does nothing if we're in MODE_IGNORE or if the data is whitespace
 467          * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 468          * data in the other modes).
 469          *
 470          * As an example, this happens when we encounter XMP like:
 471          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 472          * and are processing the 0/10 bit.
 473          *
 474          * @param XMLParser $parser XMLParser reference to the xml parser
 475          * @param string $data Character data
 476          * @throws RuntimeException On invalid data
 477          */
 478         function char( $parser, $data ) {
 479
 480                 $data = trim( $data );
 481                 if ( trim( $data ) === "" ) {
 482                         return;
 483                 }
 484
 485                 if ( !isset( $this->mode[0] ) ) {
 486                         throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
 487                 }
 488
 489                 if ( $this->mode[0] === self::MODE_IGNORE ) {
 490                         return;
 491                 }
 492
 493                 if ( $this->mode[0] !== self::MODE_SIMPLE
 494                         && $this->mode[0] !== self::MODE_QDESC
 495                 ) {
 496                         throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 497                 }
 498
 499                 // to check, how does this handle w.s.
 500                 if ( $this->charContent === false ) {
 501                         $this->charContent = $data;
 502                 } else {
 503                         $this->charContent .= $data;
 504                 }
 505         }
 506
 507         /**
 508          * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
 509          * contain a doctype declaration which could contain a dos attack if we
 510          * parse it and expand internal entities (T85848).
 511          *
 512          * @param string $content xml string to check for parse safety
 513          * @return bool true if the xml is safe to parse, false otherwise
 514          */
 515         private function checkParseSafety( $content ) {
 516                 $reader = new XMLReader();
 517                 $result = null;
 518
 519                 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
 520                 // instead of using XML().
 521                 $reader->open(
 522                         'data://text/plain,' . urlencode( $content ),
 523                         null,
 524                         LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
 525                 );
 526
 527                 $oldDisable = libxml_disable_entity_loader( true );
 528                 $reset = new ScopedCallback(
 529                         'libxml_disable_entity_loader',
 530                         array( $oldDisable )
 531                 );
 532                 $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false );
 533
 534                 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
 535                 // when parsing truncated XML, which causes unit tests to fail.
 536                 MediaWiki\suppressWarnings();
 537                 while ( $reader->read() ) {
 538                         if ( $reader->nodeType === XMLReader::ELEMENT ) {
 539                                 // Reached the first element without hitting a doctype declaration
 540                                 $this->parsable = self::PARSABLE_OK;
 541                                 $result = true;
 542                                 break;
 543                         }
 544                         if ( $reader->nodeType === XMLReader::DOC_TYPE ) {
 545                                 $this->parsable = self::PARSABLE_NO;
 546                                 $result = false;
 547                                 break;
 548                         }
 549                 }
 550                 MediaWiki\restoreWarnings();
 551
 552                 if ( !is_null( $result ) ) {
 553                         return $result;
 554                 }
 555
 556                 // Reached the end of the parsable xml without finding an element
 557                 // or doctype. Buffer and try again.
 558                 $this->parsable = self::PARSABLE_BUFFERING;
 559                 $this->xmlParsableBuffer = $content;
 560                 return false;
 561         }
 562
 563         /** When we hit a closing element in MODE_IGNORE
 564          * Check to see if this is the element we started to ignore,
 565          * in which case we get out of MODE_IGNORE
 566          *
 567          * @param string $elm Namespace of element followed by a space and then tag name of element.
 568          */
 569         private function endElementModeIgnore( $elm ) {
 570                 if ( $this->curItem[0] === $elm ) {
 571                         array_shift( $this->curItem );
 572                         array_shift( $this->mode );
 573                 }
 574         }
 575
 576         /**
 577          * Hit a closing element when in MODE_SIMPLE.
 578          * This generally means that we finished processing a
 579          * property value, and now have to save the result to the
 580          * results array
 581          *
 582          * For example, when processing:
 583          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 584          * this deals with when we hit </exif:DigitalZoomRatio>.
 585          *
 586          * Or it could be if we hit the end element of a property
 587          * of a compound data structure (like a member of an array).
 588          *
 589          * @param string $elm Namespace, space, and tag name.
 590          */
 591         private function endElementModeSimple( $elm ) {
 592                 if ( $this->charContent !== false ) {
 593                         if ( $this->processingArray ) {
 594                                 // if we're processing an array, use the original element
 595                                 // name instead of rdf:li.
 596                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 597                         } else {
 598                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 599                         }
 600                         $this->saveValue( $ns, $tag, $this->charContent );
 601
 602                         $this->charContent = false; // reset
 603                 }
 604                 array_shift( $this->curItem );
 605                 array_shift( $this->mode );
 606         }
 607
 608         /**
 609          * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 610          * generally means we've finished processing a nested structure.
 611          * resets some internal variables to indicate that.
 612          *
 613          * Note this means we hit the closing element not the "</rdf:Seq>".
 614          *
 615          * @par For example, when processing:
 616          * @code{,xml}
 617          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 618          *   </rdf:Seq> </exif:ISOSpeedRatings>
 619          * @endcode
 620          *
 621          * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
 622          *
 623          * @param string $elm Namespace . space . tag name.
 624          * @throws RuntimeException
 625          */
 626         private function endElementNested( $elm ) {
 627
 628                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 629                    in which case it could also be rdf:Description */
 630                 if ( $this->curItem[0] !== $elm
 631                         && !( $elm === self::NS_RDF . ' Description'
 632                                 && $this->mode[0] === self::MODE_STRUCT )
 633                 ) {
 634                         throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
 635                                 $this->curItem[0] . '>' );
 636                 }
 637
 638                 // Validate structures.
 639                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 640                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 641                         $info =& $this->items[$ns][$tag];
 642                         $finalName = isset( $info['map_name'] )
 643                                 ? $info['map_name'] : $tag;
 644
 645                         if ( is_array( $info['validate'] ) ) {
 646                                 $validate = $info['validate'];
 647                         } else {
 648                                 $validator = new XMPValidate( $this->logger );
 649                                 $validate = array( $validator, $info['validate'] );
 650                         }
 651
 652                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 653                                 // This can happen if all the members of the struct failed validation.
 654                                 $this->logger->debug( __METHOD__ . " <$ns:$tag> has no valid members." );
 655                         } elseif ( is_callable( $validate ) ) {
 656                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 657                                 call_user_func_array( $validate, array( $info, &$val, false ) );
 658                                 if ( is_null( $val ) ) {
 659                                         // the idea being the validation function will unset the variable if
 660                                         // its invalid.
 661                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
 662                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 663                                 }
 664                         } else {
 665                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
 666                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 667                         }
 668                 }
 669
 670                 array_shift( $this->curItem );
 671                 array_shift( $this->mode );
 672                 $this->ancestorStruct = false;
 673                 $this->processingArray = false;
 674                 $this->itemLang = false;
 675         }
 676
 677         /**
 678          * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 679          * Add information about what type of element this is.
 680          *
 681          * Note we still have to hit the outer "</property>"
 682          *
 683          * @par For example, when processing:
 684          * @code{,xml}
 685          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 686          *   </rdf:Seq> </exif:ISOSpeedRatings>
 687          * @endcode
 688          *
 689          * This method is called when we hit the "</rdf:Seq>".
 690          * (For comparison, we call endElementModeSimple when we
 691          * hit the "</rdf:li>")
 692          *
 693          * @param string $elm Namespace . ' ' . element name
 694          * @throws RuntimeException
 695          */
 696         private function endElementModeLi( $elm ) {
 697
 698                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 699                 $info = $this->items[$ns][$tag];
 700                 $finalName = isset( $info['map_name'] )
 701                         ? $info['map_name'] : $tag;
 702
 703                 array_shift( $this->mode );
 704
 705                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 706                         $this->logger->debug( __METHOD__ . " Empty compund element $finalName." );
 707
 708                         return;
 709                 }
 710
 711                 if ( $elm === self::NS_RDF . ' Seq' ) {
 712                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 713                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 714                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 715                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 716                         // extra if needed as you could theoretically have a non-language alt.
 717                         if ( $info['mode'] === self::MODE_LANG ) {
 718                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 719                         }
 720                 } else {
 721                         throw new RuntimeException( __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
 722                 }
 723         }
 724
 725         /**
 726          * End element while in MODE_QDESC
 727          * mostly when ending an element when we have a simple value
 728          * that has qualifiers.
 729          *
 730          * Qualifiers aren't all that common, and we don't do anything
 731          * with them.
 732          *
 733          * @param string $elm Namespace and element
 734          */
 735         private function endElementModeQDesc( $elm ) {
 736
 737                 if ( $elm === self::NS_RDF . ' value' ) {
 738                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 739                         $this->saveValue( $ns, $tag, $this->charContent );
 740
 741                         return;
 742                 } else {
 743                         array_shift( $this->mode );
 744                         array_shift( $this->curItem );
 745                 }
 746         }
 747
 748         /**
 749          * Handler for hitting a closing element.
 750          *
 751          * generally just calls a helper function depending on what
 752          * mode we're in.
 753          *
 754          * Ignores the outer wrapping elements that are optional in
 755          * xmp and have no meaning.
 756          *
 757          * @param XMLParser $parser
 758          * @param string $elm Namespace . ' ' . element name
 759          * @throws RuntimeException
 760          */
 761         function endElement( $parser, $elm ) {
 762                 if ( $elm === ( self::NS_RDF . ' RDF' )
 763                         || $elm === 'adobe:ns:meta/ xmpmeta'
 764                         || $elm === 'adobe:ns:meta/ xapmeta'
 765                 ) {
 766                         // ignore these.
 767                         return;
 768                 }
 769
 770                 if ( $elm === self::NS_RDF . ' type' ) {
 771                         // these aren't really supported properly yet.
 772                         // However, it appears they almost never used.
 773                         $this->logger->info( __METHOD__ . ' encountered <rdf:type>' );
 774                 }
 775
 776                 if ( strpos( $elm, ' ' ) === false ) {
 777                         // This probably shouldn't happen.
 778                         // However, there is a bug in an adobe product
 779                         // that forgets the namespace on some things.
 780                         // (Luckily they are unimportant things).
 781                         $this->logger->info( __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 782
 783                         return;
 784                 }
 785
 786                 if ( count( $this->mode[0] ) === 0 ) {
 787                         // This should never ever happen and means
 788                         // there is a pretty major bug in this class.
 789                         throw new RuntimeException( 'Encountered end element with no mode' );
 790                 }
 791
 792                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 793                         // just to be paranoid. Should always have a curItem, except for initially
 794                         // (aka during MODE_INITAL).
 795                         throw new RuntimeException( "Hit end element </$elm> but no curItem" );
 796                 }
 797
 798                 switch ( $this->mode[0] ) {
 799                         case self::MODE_IGNORE:
 800                                 $this->endElementModeIgnore( $elm );
 801                                 break;
 802                         case self::MODE_SIMPLE:
 803                                 $this->endElementModeSimple( $elm );
 804                                 break;
 805                         case self::MODE_STRUCT:
 806                         case self::MODE_SEQ:
 807                         case self::MODE_BAG:
 808                         case self::MODE_LANG:
 809                         case self::MODE_BAGSTRUCT:
 810                                 $this->endElementNested( $elm );
 811                                 break;
 812                         case self::MODE_INITIAL:
 813                                 if ( $elm === self::NS_RDF . ' Description' ) {
 814                                         array_shift( $this->mode );
 815                                 } else {
 816                                         throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
 817                                 }
 818                                 break;
 819                         case self::MODE_LI:
 820                         case self::MODE_LI_LANG:
 821                                 $this->endElementModeLi( $elm );
 822                                 break;
 823                         case self::MODE_QDESC:
 824                                 $this->endElementModeQDesc( $elm );
 825                                 break;
 826                         default:
 827                                 $this->logger->warning( __METHOD__ . " no mode (elm = $elm)" );
 828                                 break;
 829                 }
 830         }
 831
 832         /**
 833          * Hit an opening element while in MODE_IGNORE
 834          *
 835          * XMP is extensible, so ignore any tag we don't understand.
 836          *
 837          * Mostly ignores, unless we encounter the element that we are ignoring.
 838          * in which case we add it to the item stack, so we can ignore things
 839          * that are nested, correctly.
 840          *
 841          * @param string $elm Namespace . ' ' . tag name
 842          */
 843         private function startElementModeIgnore( $elm ) {
 844                 if ( $elm === $this->curItem[0] ) {
 845                         array_unshift( $this->curItem, $elm );
 846                         array_unshift( $this->mode, self::MODE_IGNORE );
 847                 }
 848         }
 849
 850         /**
 851          *  Start element in MODE_BAG (unordered array)
 852          * this should always be <rdf:Bag>
 853          *
 854          * @param string $elm Namespace . ' ' . tag
 855          * @throws RuntimeException If we have an element that's not <rdf:Bag>
 856          */
 857         private function startElementModeBag( $elm ) {
 858                 if ( $elm === self::NS_RDF . ' Bag' ) {
 859                         array_unshift( $this->mode, self::MODE_LI );
 860                 } else {
 861                         throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
 862                 }
 863         }
 864
 865         /**
 866          * Start element in MODE_SEQ (ordered array)
 867          * this should always be <rdf:Seq>
 868          *
 869          * @param string $elm Namespace . ' ' . tag
 870          * @throws RuntimeException If we have an element that's not <rdf:Seq>
 871          */
 872         private function startElementModeSeq( $elm ) {
 873                 if ( $elm === self::NS_RDF . ' Seq' ) {
 874                         array_unshift( $this->mode, self::MODE_LI );
 875                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 876                         # bug 27105
 877                         $this->logger->info( __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 878                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 879                         array_unshift( $this->mode, self::MODE_LI );
 880                 } else {
 881                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 882                 }
 883         }
 884
 885         /**
 886          * Start element in MODE_LANG (language alternative)
 887          * this should always be <rdf:Alt>
 888          *
 889          * This tag tends to be used for metadata like describe this
 890          * picture, which can be translated into multiple languages.
 891          *
 892          * XMP supports non-linguistic alternative selections,
 893          * which are really only used for thumbnails, which
 894          * we don't care about.
 895          *
 896          * @param string $elm Namespace . ' ' . tag
 897          * @throws RuntimeException If we have an element that's not <rdf:Alt>
 898          */
 899         private function startElementModeLang( $elm ) {
 900                 if ( $elm === self::NS_RDF . ' Alt' ) {
 901                         array_unshift( $this->mode, self::MODE_LI_LANG );
 902                 } else {
 903                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 904                 }
 905         }
 906
 907         /**
 908          * Handle an opening element when in MODE_SIMPLE
 909          *
 910          * This should not happen often. This is for if a simple element
 911          * already opened has a child element. Could happen for a
 912          * qualified element.
 913          *
 914          * For example:
 915          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 916          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 917          *   </exif:DigitalZoomRatio>
 918          *
 919          * This method is called when processing the <rdf:Description> element
 920          *
 921          * @param string $elm Namespace and tag names separated by space.
 922          * @param array $attribs Attributes of the element.
 923          * @throws RuntimeException
 924          */
 925         private function startElementModeSimple( $elm, $attribs ) {
 926                 if ( $elm === self::NS_RDF . ' Description' ) {
 927                         // If this value has qualifiers
 928                         array_unshift( $this->mode, self::MODE_QDESC );
 929                         array_unshift( $this->curItem, $this->curItem[0] );
 930
 931                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 932                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 933                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 934                         }
 935                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 936                         // This should not be here.
 937                         throw new RuntimeException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 938                 } else {
 939                         // something else we don't recognize, like a qualifier maybe.
 940                         $this->logger->info( __METHOD__ .
 941                                 " Encountered element <$elm> where only expecting character data as value of " .
 942                                 $this->curItem[0] );
 943                         array_unshift( $this->mode, self::MODE_IGNORE );
 944                         array_unshift( $this->curItem, $elm );
 945                 }
 946         }
 947
 948         /**
 949          * Start an element when in MODE_QDESC.
 950          * This generally happens when a simple element has an inner
 951          * rdf:Description to hold qualifier elements.
 952          *
 953          * For example in:
 954          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 955          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 956          *   </exif:DigitalZoomRatio>
 957          * Called when processing the <rdf:value> or <foo:someQualifier>.
 958          *
 959          * @param string $elm Namespace and tag name separated by a space.
 960          *
 961          */
 962         private function startElementModeQDesc( $elm ) {
 963                 if ( $elm === self::NS_RDF . ' value' ) {
 964                         return; // do nothing
 965                 } else {
 966                         // otherwise its a qualifier, which we ignore
 967                         array_unshift( $this->mode, self::MODE_IGNORE );
 968                         array_unshift( $this->curItem, $elm );
 969                 }
 970         }
 971
 972         /**
 973          * Starting an element when in MODE_INITIAL
 974          * This usually happens when we hit an element inside
 975          * the outer rdf:Description
 976          *
 977          * This is generally where most properties start.
 978          *
 979          * @param string $ns Namespace
 980          * @param string $tag Tag name (without namespace prefix)
 981          * @param array $attribs Array of attributes
 982          * @throws RuntimeException
 983          */
 984         private function startElementModeInitial( $ns, $tag, $attribs ) {
 985                 if ( $ns !== self::NS_RDF ) {
 986
 987                         if ( isset( $this->items[$ns][$tag] ) ) {
 988                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
 989                                         // If this element is supposed to appear only as
 990                                         // a child of a structure, but appears here (not as
 991                                         // a child of a struct), then something weird is
 992                                         // happening, so ignore this element and its children.
 993
 994                                         $this->logger->warning( "Encountered <$ns:$tag> outside"
 995                                                 . " of its expected parent. Ignoring." );
 996
 997                                         array_unshift( $this->mode, self::MODE_IGNORE );
 998                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
 999
1000                                         return;
1001                                 }
1002                                 $mode = $this->items[$ns][$tag]['mode'];
1003                                 array_unshift( $this->mode, $mode );
1004                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1005                                 if ( $mode === self::MODE_STRUCT ) {
1006                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
1007                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
1008                                 }
1009                                 if ( $this->charContent !== false ) {
1010                                         // Something weird.
1011                                         // Should not happen in valid XMP.
1012                                         throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1013                                 }
1014                         } else {
1015                                 // This element is not on our list of allowed elements so ignore.
1016                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1017                                 array_unshift( $this->mode, self::MODE_IGNORE );
1018                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1019
1020                                 return;
1021                         }
1022                 }
1023                 // process attributes
1024                 $this->doAttribs( $attribs );
1025         }
1026
1027         /**
1028          * Hit an opening element when in a Struct (MODE_STRUCT)
1029          * This is generally for fields of a compound property.
1030          *
1031          * Example of a struct (abbreviated; flash has more properties):
1032          *
1033          * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1034          *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1035          *
1036          * or:
1037          *
1038          * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1039          *  <exif:Mode>1</exif:Mode></exif:Flash>
1040          *
1041          * @param string $ns Namespace
1042          * @param string $tag Tag name (no ns)
1043          * @param array $attribs Array of attribs w/ values.
1044          * @throws RuntimeException
1045          */
1046         private function startElementModeStruct( $ns, $tag, $attribs ) {
1047                 if ( $ns !== self::NS_RDF ) {
1048
1049                         if ( isset( $this->items[$ns][$tag] ) ) {
1050                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
1051                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] )
1052                                 ) {
1053                                         // This assumes that we don't have inter-namespace nesting
1054                                         // which we don't in all the properties we're interested in.
1055                                         throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1056                                                 . "> where it is not allowed." );
1057                                 }
1058                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
1059                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1060                                 if ( $this->charContent !== false ) {
1061                                         // Something weird.
1062                                         // Should not happen in valid XMP.
1063                                         throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1064                                                 $this->charContent . ")." );
1065                                 }
1066                         } else {
1067                                 array_unshift( $this->mode, self::MODE_IGNORE );
1068                                 array_unshift( $this->curItem, $elm );
1069
1070                                 return;
1071                         }
1072                 }
1073
1074                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
1075                         $this->doAttribs( $attribs );
1076                         array_unshift( $this->mode, self::MODE_STRUCT );
1077                         array_unshift( $this->curItem, $this->curItem[0] );
1078                 }
1079         }
1080
1081         /**
1082          * opening element in MODE_LI
1083          * process elements of arrays.
1084          *
1085          * Example:
1086          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1087          *   </rdf:Seq> </exif:ISOSpeedRatings>
1088          * This method is called when we hit the <rdf:li> element.
1089          *
1090          * @param string $elm Namespace . ' ' . tagname
1091          * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1092          * @throws RuntimeException If gets a tag other than <rdf:li>
1093          */
1094         private function startElementModeLi( $elm, $attribs ) {
1095                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
1096                         throw new RuntimeException( "<rdf:li> expected but got $elm." );
1097                 }
1098
1099                 if ( !isset( $this->mode[1] ) ) {
1100                         // This should never ever ever happen. Checking for it
1101                         // to be paranoid.
1102                         throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1103                 }
1104
1105                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
1106                         // This list item contains a compound (STRUCT) value.
1107                         array_unshift( $this->mode, self::MODE_STRUCT );
1108                         array_unshift( $this->curItem, $elm );
1109                         $this->processingArray = true;
1110
1111                         if ( !isset( $this->curItem[1] ) ) {
1112                                 // be paranoid.
1113                                 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1114                         }
1115                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
1116                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
1117                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
1118
1119                         $this->doAttribs( $attribs );
1120                 } else {
1121                         // Normal BAG or SEQ containing simple values.
1122                         array_unshift( $this->mode, self::MODE_SIMPLE );
1123                         // need to add curItem[0] on again since one is for the specific item
1124                         // and one is for the entire group.
1125                         array_unshift( $this->curItem, $this->curItem[0] );
1126                         $this->processingArray = true;
1127                 }
1128         }
1129
1130         /**
1131          * Opening element in MODE_LI_LANG.
1132          * process elements of language alternatives
1133          *
1134          * Example:
1135          * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1136          *  </rdf:li> </rdf:Alt> </dc:title>
1137          *
1138          * This method is called when we hit the <rdf:li> element.
1139          *
1140          * @param string $elm Namespace . ' ' . tag
1141          * @param array $attribs Array of elements (most importantly xml:lang)
1142          * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1143          */
1144         private function startElementModeLiLang( $elm, $attribs ) {
1145                 if ( $elm !== self::NS_RDF . ' li' ) {
1146                         throw new RuntimeException( __METHOD__ . " <rdf:li> expected but got $elm." );
1147                 }
1148                 if ( !isset( $attribs[self::NS_XML . ' lang'] )
1149                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] )
1150                 ) {
1151                         throw new RuntimeException( __METHOD__
1152                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1153                 }
1154
1155                 // Lang is case-insensitive.
1156                 $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] );
1157
1158                 // need to add curItem[0] on again since one is for the specific item
1159                 // and one is for the entire group.
1160                 array_unshift( $this->curItem, $this->curItem[0] );
1161                 array_unshift( $this->mode, self::MODE_SIMPLE );
1162                 $this->processingArray = true;
1163         }
1164
1165         /**
1166          * Hits an opening element.
1167          * Generally just calls a helper based on what MODE we're in.
1168          * Also does some initial set up for the wrapper element
1169          *
1170          * @param XMLParser $parser
1171          * @param string $elm Namespace "<space>" element
1172          * @param array $attribs Attribute name => value
1173          * @throws RuntimeException
1174          */
1175         function startElement( $parser, $elm, $attribs ) {
1176
1177                 if ( $elm === self::NS_RDF . ' RDF'
1178                         || $elm === 'adobe:ns:meta/ xmpmeta'
1179                         || $elm === 'adobe:ns:meta/ xapmeta'
1180                 ) {
1181                         /* ignore. */
1182                         return;
1183                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1184                         if ( count( $this->mode ) === 0 ) {
1185                                 // outer rdf:desc
1186                                 array_unshift( $this->mode, self::MODE_INITIAL );
1187                         }
1188                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1189                         // This doesn't support rdf:type properly.
1190                         // In practise I have yet to see a file that
1191                         // uses this element, however it is mentioned
1192                         // on page 25 of part 1 of the xmp standard.
1193                         //
1194                         // also it seems as if exiv2 and exiftool do not support
1195                         // this either (That or I misunderstand the standard)
1196                         $this->logger->info( __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1197                 }
1198
1199                 if ( strpos( $elm, ' ' ) === false ) {
1200                         // This probably shouldn't happen.
1201                         $this->logger->info( __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1202
1203                         return;
1204                 }
1205
1206                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1207
1208                 if ( count( $this->mode ) === 0 ) {
1209                         // This should not happen.
1210                         throw new RuntimeException( 'Error extracting XMP, '
1211                                 . "encountered <$elm> with no mode" );
1212                 }
1213
1214                 switch ( $this->mode[0] ) {
1215                         case self::MODE_IGNORE:
1216                                 $this->startElementModeIgnore( $elm );
1217                                 break;
1218                         case self::MODE_SIMPLE:
1219                                 $this->startElementModeSimple( $elm, $attribs );
1220                                 break;
1221                         case self::MODE_INITIAL:
1222                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1223                                 break;
1224                         case self::MODE_STRUCT:
1225                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1226                                 break;
1227                         case self::MODE_BAG:
1228                         case self::MODE_BAGSTRUCT:
1229                                 $this->startElementModeBag( $elm );
1230                                 break;
1231                         case self::MODE_SEQ:
1232                                 $this->startElementModeSeq( $elm );
1233                                 break;
1234                         case self::MODE_LANG:
1235                                 $this->startElementModeLang( $elm );
1236                                 break;
1237                         case self::MODE_LI_LANG:
1238                                 $this->startElementModeLiLang( $elm, $attribs );
1239                                 break;
1240                         case self::MODE_LI:
1241                                 $this->startElementModeLi( $elm, $attribs );
1242                                 break;
1243                         case self::MODE_QDESC:
1244                                 $this->startElementModeQDesc( $elm );
1245                                 break;
1246                         default:
1247                                 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode[0] );
1248                 }
1249         }
1250
1251         /**
1252          * Process attributes.
1253          * Simple values can be stored as either a tag or attribute
1254          *
1255          * Often the initial "<rdf:Description>" tag just has all the simple
1256          * properties as attributes.
1257          *
1258          * @codingStandardsIgnoreStart Long line that cannot be broken
1259          * @par Example:
1260          * @code
1261          * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1262          * @endcode
1263          * @codingStandardsIgnoreEnd
1264          *
1265          * @param array $attribs Array attribute=>value
1266          * @throws RuntimeException
1267          */
1268         private function doAttribs( $attribs ) {
1269                 // first check for rdf:parseType attribute, as that can change
1270                 // how the attributes are interperted.
1271
1272                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1273                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1274                         && $this->mode[0] === self::MODE_SIMPLE
1275                 ) {
1276                         // this is equivalent to having an inner rdf:Description
1277                         $this->mode[0] = self::MODE_QDESC;
1278                 }
1279                 foreach ( $attribs as $name => $val ) {
1280                         if ( strpos( $name, ' ' ) === false ) {
1281                                 // This shouldn't happen, but so far some old software forgets namespace
1282                                 // on rdf:about.
1283                                 $this->logger->info( __METHOD__ . ' Encountered non-namespaced attribute: '
1284                                         . " $name=\"$val\". Skipping. " );
1285                                 continue;
1286                         }
1287                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1288                         if ( $ns === self::NS_RDF ) {
1289                                 if ( $tag === 'value' || $tag === 'resource' ) {
1290                                         // resource is for url.
1291                                         // value attribute is a weird way of just putting the contents.
1292                                         $this->char( $this->xmlParser, $val );
1293                                 }
1294                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1295                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1296                                         throw new RuntimeException( __METHOD__
1297                                                 . " $ns:$tag found as attribute where not allowed" );
1298                                 }
1299                                 $this->saveValue( $ns, $tag, $val );
1300                         } else {
1301                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1302                         }
1303                 }
1304         }
1305
1306         /**
1307          * Given an extracted value, save it to results array
1308          *
1309          * note also uses $this->ancestorStruct and
1310          * $this->processingArray to determine what name to
1311          * save the value under. (in addition to $tag).
1312          *
1313          * @param string $ns Namespace of tag this is for
1314          * @param string $tag Tag name
1315          * @param string $val Value to save
1316          */
1317         private function saveValue( $ns, $tag, $val ) {
1318
1319                 $info =& $this->items[$ns][$tag];
1320                 $finalName = isset( $info['map_name'] )
1321                         ? $info['map_name'] : $tag;
1322                 if ( isset( $info['validate'] ) ) {
1323                         if ( is_array( $info['validate'] ) ) {
1324                                 $validate = $info['validate'];
1325                         } else {
1326                                 $validator = new XMPValidate( $this->logger );
1327                                 $validate = array( $validator, $info['validate'] );
1328                         }
1329
1330                         if ( is_callable( $validate ) ) {
1331                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1332                                 // the reasoning behind using &$val instead of using the return value
1333                                 // is to be consistent between here and validating structures.
1334                                 if ( is_null( $val ) ) {
1335                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
1336
1337                                         return;
1338                                 }
1339                         } else {
1340                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
1341                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1342                         }
1343                 }
1344
1345                 if ( $this->ancestorStruct && $this->processingArray ) {
1346                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1347                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1348                 } elseif ( $this->ancestorStruct ) {
1349                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1350                 } elseif ( $this->processingArray ) {
1351                         if ( $this->itemLang === false ) {
1352                                 // normal array
1353                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1354                         } else {
1355                                 // lang array.
1356                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1357                         }
1358                 } else {
1359                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1360                 }
1361         }
1362 }