includes/libs/xmp/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 use Wikimedia\ScopedCallback;
  28
  29 /**
  30  * Class for reading xmp data containing properties relevant to
  31  * images, and spitting out an array that FormatMetadata accepts.
  32  *
  33  * Note, this is not meant to recognize every possible thing you can
  34  * encode in XMP. It should recognize all the properties we want.
  35  * For example it doesn't have support for structures with multiple
  36  * nesting levels, as none of the properties we're supporting use that
  37  * feature. If it comes across properties it doesn't recognize, it should
  38  * ignore them.
  39  *
  40  * The public methods one would call in this class are
  41  * - parse( $content )
  42  *    Reads in xmp content.
  43  *    Can potentially be called multiple times with partial data each time.
  44  * - parseExtended( $content )
  45  *    Reads XMPExtended blocks (jpeg files only).
  46  * - getResults
  47  *    Outputs a results array.
  48  *
  49  * Note XMP kind of looks like rdf. They are not the same thing - XMP is
  50  * encoded as a specific subset of rdf. This class can read XMP. It cannot
  51  * read rdf.
  52  *
  53  */
  54 class XMPReader implements LoggerAwareInterface {
  55         /** @var array XMP item configuration array */
  56         protected $items;
  57
  58         /** @var array Array to hold the current element (and previous element, and so on) */
  59         private $curItem = [];
  60
  61         /** @var bool|string The structure name when processing nested structures. */
  62         private $ancestorStruct = false;
  63
  64         /** @var bool|string Temporary holder for character data that appears in xmp doc. */
  65         private $charContent = false;
  66
  67         /** @var array Stores the state the xmpreader is in (see MODE_FOO constants) */
  68         private $mode = [];
  69
  70         /** @var array Array to hold results */
  71         private $results = [];
  72
  73         /** @var bool If we're doing a seq or bag. */
  74         private $processingArray = false;
  75
  76         /** @var bool|string Used for lang alts only */
  77         private $itemLang = false;
  78
  79         /** @var resource A resource handle for the XML parser */
  80         private $xmlParser;
  81
  82         /** @var bool|string Character set like 'UTF-8' */
  83         private $charset = false;
  84
  85         /** @var int */
  86         private $extendedXMPOffset = 0;
  87
  88         /** @var int Flag determining if the XMP is safe to parse **/
  89         private $parsable = 0;
  90
  91         /** @var string Buffer of XML to parse **/
  92         private $xmlParsableBuffer = '';
  93
  94         /**
  95          * These are various mode constants.
  96          * they are used to figure out what to do
  97          * with an element when its encountered.
  98          *
  99          * For example, MODE_IGNORE is used when processing
 100          * a property we're not interested in. So if a new
 101          * element pops up when we're in that mode, we ignore it.
 102          */
 103         const MODE_INITIAL = 0;
 104         const MODE_IGNORE = 1;
 105         const MODE_LI = 2;
 106         const MODE_LI_LANG = 3;
 107         const MODE_QDESC = 4;
 108
 109         // The following MODE constants are also used in the
 110         // $items array to denote what type of property the item is.
 111         const MODE_SIMPLE = 10;
 112         const MODE_STRUCT = 11; // structure (associative array)
 113         const MODE_SEQ = 12; // ordered list
 114         const MODE_BAG = 13; // unordered list
 115         const MODE_LANG = 14;
 116         const MODE_ALT = 15; // non-language alt. Currently not implemented, and not needed atm.
 117         const MODE_BAGSTRUCT = 16; // A BAG of Structs.
 118
 119         const NS_RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
 120         const NS_XML = 'http://www.w3.org/XML/1998/namespace';
 121
 122         // States used while determining if XML is safe to parse
 123         const PARSABLE_UNKNOWN = 0;
 124         const PARSABLE_OK = 1;
 125         const PARSABLE_BUFFERING = 2;
 126         const PARSABLE_NO = 3;
 127
 128         /**
 129          * @var LoggerInterface
 130          */
 131         private $logger;
 132
 133         /**
 134          * Constructor.
 135          *
 136          * Primary job is to initialize the XMLParser
 137          */
 138         function __construct( LoggerInterface $logger = null ) {
 139
 140                 if ( !function_exists( 'xml_parser_create_ns' ) ) {
 141                         // this should already be checked by this point
 142                         throw new RuntimeException( 'XMP support requires XML Parser' );
 143                 }
 144                 if ( $logger ) {
 145                         $this->setLogger( $logger );
 146                 } else {
 147                         $this->setLogger( new NullLogger() );
 148                 }
 149
 150                 $this->items = XMPInfo::getItems();
 151
 152                 $this->resetXMLParser();
 153         }
 154
 155         public function setLogger( LoggerInterface $logger ) {
 156                 $this->logger = $logger;
 157         }
 158
 159         /**
 160          * free the XML parser.
 161          *
 162          * @note It is unclear to me if we really need to do this ourselves
 163          *  or if php garbage collection will automatically free the xmlParser
 164          *  when it is no longer needed.
 165          */
 166         private function destroyXMLParser() {
 167                 if ( $this->xmlParser ) {
 168                         xml_parser_free( $this->xmlParser );
 169                         $this->xmlParser = null;
 170                 }
 171         }
 172
 173         /**
 174          * Main use is if a single item has multiple xmp documents describing it.
 175          * For example in jpeg's with extendedXMP
 176          */
 177         private function resetXMLParser() {
 178
 179                 $this->destroyXMLParser();
 180
 181                 $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' );
 182                 xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 );
 183                 xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 );
 184
 185                 xml_set_element_handler( $this->xmlParser,
 186                         [ $this, 'startElement' ],
 187                         [ $this, 'endElement' ] );
 188
 189                 xml_set_character_data_handler( $this->xmlParser, [ $this, 'char' ] );
 190
 191                 $this->parsable = self::PARSABLE_UNKNOWN;
 192                 $this->xmlParsableBuffer = '';
 193         }
 194
 195         /**
 196          * Check if this instance supports using this class
 197          */
 198         public static function isSupported() {
 199                 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
 200         }
 201
 202         /** Get the result array. Do some post-processing before returning
 203          * the array, and transform any metadata that is special-cased.
 204          *
 205          * @return array Array of results as an array of arrays suitable for
 206          *    FormatMetadata::getFormattedData().
 207          */
 208         public function getResults() {
 209                 // xmp-special is for metadata that affects how stuff
 210                 // is extracted. For example xmpNote:HasExtendedXMP.
 211
 212                 // It is also used to handle photoshop:AuthorsPosition
 213                 // which is weird and really part of another property,
 214                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 215                 // The location fields also use it.
 216
 217                 $data = $this->results;
 218
 219                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 220                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 221                         && isset( $data['xmp-general']['Artist'][0] )
 222                 ) {
 223                         // Note, if there is more than one creator,
 224                         // this only applies to first. This also will
 225                         // only apply to the dc:Creator prop, not the
 226                         // exif:Artist prop.
 227
 228                         $data['xmp-general']['Artist'][0] =
 229                                 $data['xmp-special']['AuthorsPosition'] . ', '
 230                                 . $data['xmp-general']['Artist'][0];
 231                 }
 232
 233                 // Go through the LocationShown and LocationCreated
 234                 // changing it to the non-hierarchal form used by
 235                 // the other location fields.
 236
 237                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 238                         && is_array( $data['xmp-special']['LocationShown'][0] )
 239                 ) {
 240                         // the is_array is just paranoia. It should always
 241                         // be an array.
 242                         foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
 243                                 if ( !is_array( $loc ) ) {
 244                                         // To avoid copying over the _type meta-fields.
 245                                         continue;
 246                                 }
 247                                 foreach ( $loc as $field => $val ) {
 248                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 249                                 }
 250                         }
 251                 }
 252                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 253                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 254                 ) {
 255                         // the is_array is just paranoia. It should always
 256                         // be an array.
 257                         foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
 258                                 if ( !is_array( $loc ) ) {
 259                                         // To avoid copying over the _type meta-fields.
 260                                         continue;
 261                                 }
 262                                 foreach ( $loc as $field => $val ) {
 263                                         $data['xmp-general'][$field . 'Created'][] = $val;
 264                                 }
 265                         }
 266                 }
 267
 268                 // We don't want to return the special values, since they're
 269                 // special and not info to be stored about the file.
 270                 unset( $data['xmp-special'] );
 271
 272                 // Convert GPSAltitude to negative if below sea level.
 273                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
 274                         && isset( $data['xmp-exif']['GPSAltitude'] )
 275                 ) {
 276
 277                         // Must convert to a real before multiplying by -1
 278                         // XMPValidate guarantees there will always be a '/' in this value.
 279                         list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
 280                         $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
 281
 282                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
 283                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 284                         }
 285                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 286                 }
 287
 288                 return $data;
 289         }
 290
 291         /**
 292          * Main function to call to parse XMP. Use getResults to
 293          * get results.
 294          *
 295          * Also catches any errors during processing, writes them to
 296          * debug log, blanks result array and returns false.
 297          *
 298          * @param string $content XMP data
 299          * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
 300          * @throws RuntimeException
 301          * @return bool Success.
 302          */
 303         public function parse( $content, $allOfIt = true ) {
 304                 if ( !$this->xmlParser ) {
 305                         $this->resetXMLParser();
 306                 }
 307                 try {
 308
 309                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 310                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 311                         if ( !$this->charset ) {
 312                                 $bom = [];
 313                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 314                                         $content, $bom )
 315                                 ) {
 316                                         switch ( $bom[0] ) {
 317                                                 case "\xFE\xFF":
 318                                                         $this->charset = 'UTF-16BE';
 319                                                         break;
 320                                                 case "\xFF\xFE":
 321                                                         $this->charset = 'UTF-16LE';
 322                                                         break;
 323                                                 case "\x00\x00\xFE\xFF":
 324                                                         $this->charset = 'UTF-32BE';
 325                                                         break;
 326                                                 case "\xFF\xFE\x00\x00":
 327                                                         $this->charset = 'UTF-32LE';
 328                                                         break;
 329                                                 case "\xEF\xBB\xBF":
 330                                                         $this->charset = 'UTF-8';
 331                                                         break;
 332                                                 default:
 333                                                         // this should be impossible to get to
 334                                                         throw new RuntimeException( "Invalid BOM" );
 335                                         }
 336                                 } else {
 337                                         // standard specifically says, if no bom assume utf-8
 338                                         $this->charset = 'UTF-8';
 339                                 }
 340                         }
 341                         if ( $this->charset !== 'UTF-8' ) {
 342                                 // don't convert if already utf-8
 343                                 MediaWiki\suppressWarnings();
 344                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 345                                 MediaWiki\restoreWarnings();
 346                         }
 347
 348                         // Ensure the XMP block does not have an xml doctype declaration, which
 349                         // could declare entities unsafe to parse with xml_parse (T85848/T71210).
 350                         if ( $this->parsable !== self::PARSABLE_OK ) {
 351                                 if ( $this->parsable === self::PARSABLE_NO ) {
 352                                         throw new RuntimeException( 'Unsafe doctype declaration in XML.' );
 353                                 }
 354
 355                                 $content = $this->xmlParsableBuffer . $content;
 356                                 if ( !$this->checkParseSafety( $content ) ) {
 357                                         if ( !$allOfIt && $this->parsable !== self::PARSABLE_NO ) {
 358                                                 // parse wasn't Unsuccessful yet, so return true
 359                                                 // in this case.
 360                                                 return true;
 361                                         }
 362                                         $msg = ( $this->parsable === self::PARSABLE_NO ) ?
 363                                                 'Unsafe doctype declaration in XML.' :
 364                                                 'No root element found in XML.';
 365                                         throw new RuntimeException( $msg );
 366                                 }
 367                         }
 368
 369                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 370                         if ( !$ok ) {
 371                                 $code = xml_get_error_code( $this->xmlParser );
 372                                 $error = xml_error_string( $code );
 373                                 $line = xml_get_current_line_number( $this->xmlParser );
 374                                 $col = xml_get_current_column_number( $this->xmlParser );
 375                                 $offset = xml_get_current_byte_index( $this->xmlParser );
 376
 377                                 $this->logger->warning(
 378                                         '{method} : Error reading XMP content: {error} ' .
 379                                         '(line: {line} column: {column} byte offset: {offset})',
 380                                         [
 381                                                 'method' => __METHOD__,
 382                                                 'error_code' => $code,
 383                                                 'error' => $error,
 384                                                 'line' => $line,
 385                                                 'column' => $col,
 386                                                 'offset' => $offset,
 387                                                 'content' => $content,
 388                                 ] );
 389                                 $this->results = []; // blank if error.
 390                                 $this->destroyXMLParser();
 391                                 return false;
 392                         }
 393                 } catch ( Exception $e ) {
 394                         $this->logger->warning(
 395                                 '{method} Exception caught while parsing: ' . $e->getMessage(),
 396                                 [
 397                                         'method' => __METHOD__,
 398                                         'exception' => $e,
 399                                         'content' => $content,
 400                                 ]
 401                         );
 402                         $this->results = [];
 403                         return false;
 404                 }
 405                 if ( $allOfIt ) {
 406                         $this->destroyXMLParser();
 407                 }
 408
 409                 return true;
 410         }
 411
 412         /** Entry point for XMPExtended blocks in jpeg files
 413          *
 414          * @todo In serious need of testing
 415          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 416          * @param string $content XMPExtended block minus the namespace signature
 417          * @return bool If it succeeded.
 418          */
 419         public function parseExtended( $content ) {
 420                 // @todo FIXME: This is untested. Hard to find example files
 421                 // or programs that make such files..
 422                 $guid = substr( $content, 0, 32 );
 423                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 424                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid
 425                 ) {
 426                         $this->logger->info( __METHOD__ .
 427                                 " Ignoring XMPExtended block due to wrong guid (guid= '$guid')" );
 428
 429                         return false;
 430                 }
 431                 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 432
 433                 if ( !$len ||
 434                         $len['length'] < 4 ||
 435                         $len['offset'] < 0 ||
 436                         $len['offset'] > $len['length']
 437                 ) {
 438                         $this->logger->info(
 439                                 __METHOD__ . 'Error reading extended XMP block, invalid length or offset.'
 440                         );
 441
 442                         return false;
 443                 }
 444
 445                 // we're not very robust here. we should accept it in the wrong order.
 446                 // To quote the XMP standard:
 447                 // "A JPEG writer should write the ExtendedXMP marker segments in order,
 448                 // immediately following the StandardXMP. However, the JPEG standard
 449                 // does not require preservation of marker segment order. A robust JPEG
 450                 // reader should tolerate the marker segments in any order."
 451                 // On the other hand, the probability that an image will have more than
 452                 // 128k of metadata is rather low... so the probability that it will have
 453                 // > 128k, and be in the wrong order is very low...
 454
 455                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 456                         $this->logger->info( __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 457                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')' );
 458
 459                         return false;
 460                 }
 461
 462                 if ( $len['offset'] === 0 ) {
 463                         // if we're starting the extended block, we've probably already
 464                         // done the XMPStandard block, so reset.
 465                         $this->resetXMLParser();
 466                 }
 467
 468                 $this->extendedXMPOffset += $len['length'];
 469
 470                 $actualContent = substr( $content, 40 );
 471
 472                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 473                         $atEnd = true;
 474                 } else {
 475                         $atEnd = false;
 476                 }
 477
 478                 $this->logger->debug( __METHOD__ . 'Parsing a XMPExtended block' );
 479
 480                 return $this->parse( $actualContent, $atEnd );
 481         }
 482
 483         /**
 484          * Character data handler
 485          * Called whenever character data is found in the xmp document.
 486          *
 487          * does nothing if we're in MODE_IGNORE or if the data is whitespace
 488          * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 489          * data in the other modes).
 490          *
 491          * As an example, this happens when we encounter XMP like:
 492          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 493          * and are processing the 0/10 bit.
 494          *
 495          * @param XMLParser $parser XMLParser reference to the xml parser
 496          * @param string $data Character data
 497          * @throws RuntimeException On invalid data
 498          */
 499         function char( $parser, $data ) {
 500
 501                 $data = trim( $data );
 502                 if ( trim( $data ) === "" ) {
 503                         return;
 504                 }
 505
 506                 if ( !isset( $this->mode[0] ) ) {
 507                         throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
 508                 }
 509
 510                 if ( $this->mode[0] === self::MODE_IGNORE ) {
 511                         return;
 512                 }
 513
 514                 if ( $this->mode[0] !== self::MODE_SIMPLE
 515                         && $this->mode[0] !== self::MODE_QDESC
 516                 ) {
 517                         throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 518                 }
 519
 520                 // to check, how does this handle w.s.
 521                 if ( $this->charContent === false ) {
 522                         $this->charContent = $data;
 523                 } else {
 524                         $this->charContent .= $data;
 525                 }
 526         }
 527
 528         /**
 529          * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
 530          * contain a doctype declaration which could contain a dos attack if we
 531          * parse it and expand internal entities (T85848).
 532          *
 533          * @param string $content xml string to check for parse safety
 534          * @return bool true if the xml is safe to parse, false otherwise
 535          */
 536         private function checkParseSafety( $content ) {
 537                 $reader = new XMLReader();
 538                 $result = null;
 539
 540                 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
 541                 // instead of using XML().
 542                 $reader->open(
 543                         'data://text/plain,' . urlencode( $content ),
 544                         null,
 545                         LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
 546                 );
 547
 548                 $oldDisable = libxml_disable_entity_loader( true );
 549                 /** @noinspection PhpUnusedLocalVariableInspection */
 550                 $reset = new ScopedCallback(
 551                         'libxml_disable_entity_loader',
 552                         [ $oldDisable ]
 553                 );
 554                 $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false );
 555
 556                 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
 557                 // when parsing truncated XML, which causes unit tests to fail.
 558                 MediaWiki\suppressWarnings();
 559                 while ( $reader->read() ) {
 560                         if ( $reader->nodeType === XMLReader::ELEMENT ) {
 561                                 // Reached the first element without hitting a doctype declaration
 562                                 $this->parsable = self::PARSABLE_OK;
 563                                 $result = true;
 564                                 break;
 565                         }
 566                         if ( $reader->nodeType === XMLReader::DOC_TYPE ) {
 567                                 $this->parsable = self::PARSABLE_NO;
 568                                 $result = false;
 569                                 break;
 570                         }
 571                 }
 572                 MediaWiki\restoreWarnings();
 573
 574                 if ( !is_null( $result ) ) {
 575                         return $result;
 576                 }
 577
 578                 // Reached the end of the parsable xml without finding an element
 579                 // or doctype. Buffer and try again.
 580                 $this->parsable = self::PARSABLE_BUFFERING;
 581                 $this->xmlParsableBuffer = $content;
 582                 return false;
 583         }
 584
 585         /** When we hit a closing element in MODE_IGNORE
 586          * Check to see if this is the element we started to ignore,
 587          * in which case we get out of MODE_IGNORE
 588          *
 589          * @param string $elm Namespace of element followed by a space and then tag name of element.
 590          */
 591         private function endElementModeIgnore( $elm ) {
 592                 if ( $this->curItem[0] === $elm ) {
 593                         array_shift( $this->curItem );
 594                         array_shift( $this->mode );
 595                 }
 596         }
 597
 598         /**
 599          * Hit a closing element when in MODE_SIMPLE.
 600          * This generally means that we finished processing a
 601          * property value, and now have to save the result to the
 602          * results array
 603          *
 604          * For example, when processing:
 605          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 606          * this deals with when we hit </exif:DigitalZoomRatio>.
 607          *
 608          * Or it could be if we hit the end element of a property
 609          * of a compound data structure (like a member of an array).
 610          *
 611          * @param string $elm Namespace, space, and tag name.
 612          */
 613         private function endElementModeSimple( $elm ) {
 614                 if ( $this->charContent !== false ) {
 615                         if ( $this->processingArray ) {
 616                                 // if we're processing an array, use the original element
 617                                 // name instead of rdf:li.
 618                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 619                         } else {
 620                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 621                         }
 622                         $this->saveValue( $ns, $tag, $this->charContent );
 623
 624                         $this->charContent = false; // reset
 625                 }
 626                 array_shift( $this->curItem );
 627                 array_shift( $this->mode );
 628         }
 629
 630         /**
 631          * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 632          * generally means we've finished processing a nested structure.
 633          * resets some internal variables to indicate that.
 634          *
 635          * Note this means we hit the closing element not the "</rdf:Seq>".
 636          *
 637          * @par For example, when processing:
 638          * @code{,xml}
 639          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 640          *   </rdf:Seq> </exif:ISOSpeedRatings>
 641          * @endcode
 642          *
 643          * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
 644          *
 645          * @param string $elm Namespace . space . tag name.
 646          * @throws RuntimeException
 647          */
 648         private function endElementNested( $elm ) {
 649
 650                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 651                    in which case it could also be rdf:Description */
 652                 if ( $this->curItem[0] !== $elm
 653                         && !( $elm === self::NS_RDF . ' Description'
 654                                 && $this->mode[0] === self::MODE_STRUCT )
 655                 ) {
 656                         throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
 657                                 $this->curItem[0] . '>' );
 658                 }
 659
 660                 // Validate structures.
 661                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 662                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 663                         $info =& $this->items[$ns][$tag];
 664                         $finalName = isset( $info['map_name'] )
 665                                 ? $info['map_name'] : $tag;
 666
 667                         if ( is_array( $info['validate'] ) ) {
 668                                 $validate = $info['validate'];
 669                         } else {
 670                                 $validator = new XMPValidate( $this->logger );
 671                                 $validate = [ $validator, $info['validate'] ];
 672                         }
 673
 674                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 675                                 // This can happen if all the members of the struct failed validation.
 676                                 $this->logger->debug( __METHOD__ . " <$ns:$tag> has no valid members." );
 677                         } elseif ( is_callable( $validate ) ) {
 678                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 679                                 call_user_func_array( $validate, [ $info, &$val, false ] );
 680                                 if ( is_null( $val ) ) {
 681                                         // the idea being the validation function will unset the variable if
 682                                         // its invalid.
 683                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
 684                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 685                                 }
 686                         } else {
 687                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
 688                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 689                         }
 690                 }
 691
 692                 array_shift( $this->curItem );
 693                 array_shift( $this->mode );
 694                 $this->ancestorStruct = false;
 695                 $this->processingArray = false;
 696                 $this->itemLang = false;
 697         }
 698
 699         /**
 700          * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 701          * Add information about what type of element this is.
 702          *
 703          * Note we still have to hit the outer "</property>"
 704          *
 705          * @par For example, when processing:
 706          * @code{,xml}
 707          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 708          *   </rdf:Seq> </exif:ISOSpeedRatings>
 709          * @endcode
 710          *
 711          * This method is called when we hit the "</rdf:Seq>".
 712          * (For comparison, we call endElementModeSimple when we
 713          * hit the "</rdf:li>")
 714          *
 715          * @param string $elm Namespace . ' ' . element name
 716          * @throws RuntimeException
 717          */
 718         private function endElementModeLi( $elm ) {
 719                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 720                 $info = $this->items[$ns][$tag];
 721                 $finalName = isset( $info['map_name'] )
 722                         ? $info['map_name'] : $tag;
 723
 724                 array_shift( $this->mode );
 725
 726                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 727                         $this->logger->debug( __METHOD__ . " Empty compund element $finalName." );
 728
 729                         return;
 730                 }
 731
 732                 if ( $elm === self::NS_RDF . ' Seq' ) {
 733                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 734                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 735                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 736                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 737                         // extra if needed as you could theoretically have a non-language alt.
 738                         if ( $info['mode'] === self::MODE_LANG ) {
 739                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 740                         }
 741                 } else {
 742                         throw new RuntimeException(
 743                                 __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm."
 744                         );
 745                 }
 746         }
 747
 748         /**
 749          * End element while in MODE_QDESC
 750          * mostly when ending an element when we have a simple value
 751          * that has qualifiers.
 752          *
 753          * Qualifiers aren't all that common, and we don't do anything
 754          * with them.
 755          *
 756          * @param string $elm Namespace and element
 757          */
 758         private function endElementModeQDesc( $elm ) {
 759
 760                 if ( $elm === self::NS_RDF . ' value' ) {
 761                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 762                         $this->saveValue( $ns, $tag, $this->charContent );
 763
 764                         return;
 765                 } else {
 766                         array_shift( $this->mode );
 767                         array_shift( $this->curItem );
 768                 }
 769         }
 770
 771         /**
 772          * Handler for hitting a closing element.
 773          *
 774          * generally just calls a helper function depending on what
 775          * mode we're in.
 776          *
 777          * Ignores the outer wrapping elements that are optional in
 778          * xmp and have no meaning.
 779          *
 780          * @param XMLParser $parser
 781          * @param string $elm Namespace . ' ' . element name
 782          * @throws RuntimeException
 783          */
 784         function endElement( $parser, $elm ) {
 785                 if ( $elm === ( self::NS_RDF . ' RDF' )
 786                         || $elm === 'adobe:ns:meta/ xmpmeta'
 787                         || $elm === 'adobe:ns:meta/ xapmeta'
 788                 ) {
 789                         // ignore these.
 790                         return;
 791                 }
 792
 793                 if ( $elm === self::NS_RDF . ' type' ) {
 794                         // these aren't really supported properly yet.
 795                         // However, it appears they almost never used.
 796                         $this->logger->info( __METHOD__ . ' encountered <rdf:type>' );
 797                 }
 798
 799                 if ( strpos( $elm, ' ' ) === false ) {
 800                         // This probably shouldn't happen.
 801                         // However, there is a bug in an adobe product
 802                         // that forgets the namespace on some things.
 803                         // (Luckily they are unimportant things).
 804                         $this->logger->info( __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 805
 806                         return;
 807                 }
 808
 809                 if ( count( $this->mode[0] ) === 0 ) {
 810                         // This should never ever happen and means
 811                         // there is a pretty major bug in this class.
 812                         throw new RuntimeException( 'Encountered end element with no mode' );
 813                 }
 814
 815                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 816                         // just to be paranoid. Should always have a curItem, except for initially
 817                         // (aka during MODE_INITAL).
 818                         throw new RuntimeException( "Hit end element </$elm> but no curItem" );
 819                 }
 820
 821                 switch ( $this->mode[0] ) {
 822                         case self::MODE_IGNORE:
 823                                 $this->endElementModeIgnore( $elm );
 824                                 break;
 825                         case self::MODE_SIMPLE:
 826                                 $this->endElementModeSimple( $elm );
 827                                 break;
 828                         case self::MODE_STRUCT:
 829                         case self::MODE_SEQ:
 830                         case self::MODE_BAG:
 831                         case self::MODE_LANG:
 832                         case self::MODE_BAGSTRUCT:
 833                                 $this->endElementNested( $elm );
 834                                 break;
 835                         case self::MODE_INITIAL:
 836                                 if ( $elm === self::NS_RDF . ' Description' ) {
 837                                         array_shift( $this->mode );
 838                                 } else {
 839                                         throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
 840                                 }
 841                                 break;
 842                         case self::MODE_LI:
 843                         case self::MODE_LI_LANG:
 844                                 $this->endElementModeLi( $elm );
 845                                 break;
 846                         case self::MODE_QDESC:
 847                                 $this->endElementModeQDesc( $elm );
 848                                 break;
 849                         default:
 850                                 $this->logger->warning( __METHOD__ . " no mode (elm = $elm)" );
 851                                 break;
 852                 }
 853         }
 854
 855         /**
 856          * Hit an opening element while in MODE_IGNORE
 857          *
 858          * XMP is extensible, so ignore any tag we don't understand.
 859          *
 860          * Mostly ignores, unless we encounter the element that we are ignoring.
 861          * in which case we add it to the item stack, so we can ignore things
 862          * that are nested, correctly.
 863          *
 864          * @param string $elm Namespace . ' ' . tag name
 865          */
 866         private function startElementModeIgnore( $elm ) {
 867                 if ( $elm === $this->curItem[0] ) {
 868                         array_unshift( $this->curItem, $elm );
 869                         array_unshift( $this->mode, self::MODE_IGNORE );
 870                 }
 871         }
 872
 873         /**
 874          *  Start element in MODE_BAG (unordered array)
 875          * this should always be <rdf:Bag>
 876          *
 877          * @param string $elm Namespace . ' ' . tag
 878          * @throws RuntimeException If we have an element that's not <rdf:Bag>
 879          */
 880         private function startElementModeBag( $elm ) {
 881                 if ( $elm === self::NS_RDF . ' Bag' ) {
 882                         array_unshift( $this->mode, self::MODE_LI );
 883                 } else {
 884                         throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
 885                 }
 886         }
 887
 888         /**
 889          * Start element in MODE_SEQ (ordered array)
 890          * this should always be <rdf:Seq>
 891          *
 892          * @param string $elm Namespace . ' ' . tag
 893          * @throws RuntimeException If we have an element that's not <rdf:Seq>
 894          */
 895         private function startElementModeSeq( $elm ) {
 896                 if ( $elm === self::NS_RDF . ' Seq' ) {
 897                         array_unshift( $this->mode, self::MODE_LI );
 898                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 899                         # bug 27105
 900                         $this->logger->info( __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 901                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 902                         array_unshift( $this->mode, self::MODE_LI );
 903                 } else {
 904                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 905                 }
 906         }
 907
 908         /**
 909          * Start element in MODE_LANG (language alternative)
 910          * this should always be <rdf:Alt>
 911          *
 912          * This tag tends to be used for metadata like describe this
 913          * picture, which can be translated into multiple languages.
 914          *
 915          * XMP supports non-linguistic alternative selections,
 916          * which are really only used for thumbnails, which
 917          * we don't care about.
 918          *
 919          * @param string $elm Namespace . ' ' . tag
 920          * @throws RuntimeException If we have an element that's not <rdf:Alt>
 921          */
 922         private function startElementModeLang( $elm ) {
 923                 if ( $elm === self::NS_RDF . ' Alt' ) {
 924                         array_unshift( $this->mode, self::MODE_LI_LANG );
 925                 } else {
 926                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 927                 }
 928         }
 929
 930         /**
 931          * Handle an opening element when in MODE_SIMPLE
 932          *
 933          * This should not happen often. This is for if a simple element
 934          * already opened has a child element. Could happen for a
 935          * qualified element.
 936          *
 937          * For example:
 938          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 939          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 940          *   </exif:DigitalZoomRatio>
 941          *
 942          * This method is called when processing the <rdf:Description> element
 943          *
 944          * @param string $elm Namespace and tag names separated by space.
 945          * @param array $attribs Attributes of the element.
 946          * @throws RuntimeException
 947          */
 948         private function startElementModeSimple( $elm, $attribs ) {
 949                 if ( $elm === self::NS_RDF . ' Description' ) {
 950                         // If this value has qualifiers
 951                         array_unshift( $this->mode, self::MODE_QDESC );
 952                         array_unshift( $this->curItem, $this->curItem[0] );
 953
 954                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 955                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 956                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 957                         }
 958                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 959                         // This should not be here.
 960                         throw new RuntimeException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 961                 } else {
 962                         // something else we don't recognize, like a qualifier maybe.
 963                         $this->logger->info( __METHOD__ .
 964                                 " Encountered element <$elm> where only expecting character data as value of " .
 965                                 $this->curItem[0] );
 966                         array_unshift( $this->mode, self::MODE_IGNORE );
 967                         array_unshift( $this->curItem, $elm );
 968                 }
 969         }
 970
 971         /**
 972          * Start an element when in MODE_QDESC.
 973          * This generally happens when a simple element has an inner
 974          * rdf:Description to hold qualifier elements.
 975          *
 976          * For example in:
 977          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 978          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 979          *   </exif:DigitalZoomRatio>
 980          * Called when processing the <rdf:value> or <foo:someQualifier>.
 981          *
 982          * @param string $elm Namespace and tag name separated by a space.
 983          *
 984          */
 985         private function startElementModeQDesc( $elm ) {
 986                 if ( $elm === self::NS_RDF . ' value' ) {
 987                         return; // do nothing
 988                 } else {
 989                         // otherwise its a qualifier, which we ignore
 990                         array_unshift( $this->mode, self::MODE_IGNORE );
 991                         array_unshift( $this->curItem, $elm );
 992                 }
 993         }
 994
 995         /**
 996          * Starting an element when in MODE_INITIAL
 997          * This usually happens when we hit an element inside
 998          * the outer rdf:Description
 999          *
1000          * This is generally where most properties start.
1001          *
1002          * @param string $ns Namespace
1003          * @param string $tag Tag name (without namespace prefix)
1004          * @param array $attribs Array of attributes
1005          * @throws RuntimeException
1006          */
1007         private function startElementModeInitial( $ns, $tag, $attribs ) {
1008                 if ( $ns !== self::NS_RDF ) {
1009
1010                         if ( isset( $this->items[$ns][$tag] ) ) {
1011                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
1012                                         // If this element is supposed to appear only as
1013                                         // a child of a structure, but appears here (not as
1014                                         // a child of a struct), then something weird is
1015                                         // happening, so ignore this element and its children.
1016
1017                                         $this->logger->warning( "Encountered <$ns:$tag> outside"
1018                                                 . " of its expected parent. Ignoring." );
1019
1020                                         array_unshift( $this->mode, self::MODE_IGNORE );
1021                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
1022
1023                                         return;
1024                                 }
1025                                 $mode = $this->items[$ns][$tag]['mode'];
1026                                 array_unshift( $this->mode, $mode );
1027                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1028                                 if ( $mode === self::MODE_STRUCT ) {
1029                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
1030                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
1031                                 }
1032                                 if ( $this->charContent !== false ) {
1033                                         // Something weird.
1034                                         // Should not happen in valid XMP.
1035                                         throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1036                                 }
1037                         } else {
1038                                 // This element is not on our list of allowed elements so ignore.
1039                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1040                                 array_unshift( $this->mode, self::MODE_IGNORE );
1041                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1042
1043                                 return;
1044                         }
1045                 }
1046                 // process attributes
1047                 $this->doAttribs( $attribs );
1048         }
1049
1050         /**
1051          * Hit an opening element when in a Struct (MODE_STRUCT)
1052          * This is generally for fields of a compound property.
1053          *
1054          * Example of a struct (abbreviated; flash has more properties):
1055          *
1056          * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1057          *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1058          *
1059          * or:
1060          *
1061          * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1062          *  <exif:Mode>1</exif:Mode></exif:Flash>
1063          *
1064          * @param string $ns Namespace
1065          * @param string $tag Tag name (no ns)
1066          * @param array $attribs Array of attribs w/ values.
1067          * @throws RuntimeException
1068          */
1069         private function startElementModeStruct( $ns, $tag, $attribs ) {
1070                 if ( $ns !== self::NS_RDF ) {
1071
1072                         if ( isset( $this->items[$ns][$tag] ) ) {
1073                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
1074                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] )
1075                                 ) {
1076                                         // This assumes that we don't have inter-namespace nesting
1077                                         // which we don't in all the properties we're interested in.
1078                                         throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1079                                                 . "> where it is not allowed." );
1080                                 }
1081                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
1082                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1083                                 if ( $this->charContent !== false ) {
1084                                         // Something weird.
1085                                         // Should not happen in valid XMP.
1086                                         throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1087                                                 $this->charContent . ")." );
1088                                 }
1089                         } else {
1090                                 array_unshift( $this->mode, self::MODE_IGNORE );
1091                                 array_unshift( $this->curItem, $elm );
1092
1093                                 return;
1094                         }
1095                 }
1096
1097                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
1098                         $this->doAttribs( $attribs );
1099                         array_unshift( $this->mode, self::MODE_STRUCT );
1100                         array_unshift( $this->curItem, $this->curItem[0] );
1101                 }
1102         }
1103
1104         /**
1105          * opening element in MODE_LI
1106          * process elements of arrays.
1107          *
1108          * Example:
1109          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1110          *   </rdf:Seq> </exif:ISOSpeedRatings>
1111          * This method is called when we hit the <rdf:li> element.
1112          *
1113          * @param string $elm Namespace . ' ' . tagname
1114          * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1115          * @throws RuntimeException If gets a tag other than <rdf:li>
1116          */
1117         private function startElementModeLi( $elm, $attribs ) {
1118                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
1119                         throw new RuntimeException( "<rdf:li> expected but got $elm." );
1120                 }
1121
1122                 if ( !isset( $this->mode[1] ) ) {
1123                         // This should never ever ever happen. Checking for it
1124                         // to be paranoid.
1125                         throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1126                 }
1127
1128                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
1129                         // This list item contains a compound (STRUCT) value.
1130                         array_unshift( $this->mode, self::MODE_STRUCT );
1131                         array_unshift( $this->curItem, $elm );
1132                         $this->processingArray = true;
1133
1134                         if ( !isset( $this->curItem[1] ) ) {
1135                                 // be paranoid.
1136                                 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1137                         }
1138                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
1139                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
1140                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
1141
1142                         $this->doAttribs( $attribs );
1143                 } else {
1144                         // Normal BAG or SEQ containing simple values.
1145                         array_unshift( $this->mode, self::MODE_SIMPLE );
1146                         // need to add curItem[0] on again since one is for the specific item
1147                         // and one is for the entire group.
1148                         array_unshift( $this->curItem, $this->curItem[0] );
1149                         $this->processingArray = true;
1150                 }
1151         }
1152
1153         /**
1154          * Opening element in MODE_LI_LANG.
1155          * process elements of language alternatives
1156          *
1157          * Example:
1158          * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1159          *  </rdf:li> </rdf:Alt> </dc:title>
1160          *
1161          * This method is called when we hit the <rdf:li> element.
1162          *
1163          * @param string $elm Namespace . ' ' . tag
1164          * @param array $attribs Array of elements (most importantly xml:lang)
1165          * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1166          */
1167         private function startElementModeLiLang( $elm, $attribs ) {
1168                 if ( $elm !== self::NS_RDF . ' li' ) {
1169                         throw new RuntimeException( __METHOD__ . " <rdf:li> expected but got $elm." );
1170                 }
1171                 if ( !isset( $attribs[self::NS_XML . ' lang'] )
1172                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] )
1173                 ) {
1174                         throw new RuntimeException( __METHOD__
1175                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1176                 }
1177
1178                 // Lang is case-insensitive.
1179                 $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] );
1180
1181                 // need to add curItem[0] on again since one is for the specific item
1182                 // and one is for the entire group.
1183                 array_unshift( $this->curItem, $this->curItem[0] );
1184                 array_unshift( $this->mode, self::MODE_SIMPLE );
1185                 $this->processingArray = true;
1186         }
1187
1188         /**
1189          * Hits an opening element.
1190          * Generally just calls a helper based on what MODE we're in.
1191          * Also does some initial set up for the wrapper element
1192          *
1193          * @param XMLParser $parser
1194          * @param string $elm Namespace "<space>" element
1195          * @param array $attribs Attribute name => value
1196          * @throws RuntimeException
1197          */
1198         function startElement( $parser, $elm, $attribs ) {
1199
1200                 if ( $elm === self::NS_RDF . ' RDF'
1201                         || $elm === 'adobe:ns:meta/ xmpmeta'
1202                         || $elm === 'adobe:ns:meta/ xapmeta'
1203                 ) {
1204                         /* ignore. */
1205                         return;
1206                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1207                         if ( count( $this->mode ) === 0 ) {
1208                                 // outer rdf:desc
1209                                 array_unshift( $this->mode, self::MODE_INITIAL );
1210                         }
1211                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1212                         // This doesn't support rdf:type properly.
1213                         // In practise I have yet to see a file that
1214                         // uses this element, however it is mentioned
1215                         // on page 25 of part 1 of the xmp standard.
1216                         // Also it seems as if exiv2 and exiftool do not support
1217                         // this either (That or I misunderstand the standard)
1218                         $this->logger->info( __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1219                 }
1220
1221                 if ( strpos( $elm, ' ' ) === false ) {
1222                         // This probably shouldn't happen.
1223                         $this->logger->info( __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1224
1225                         return;
1226                 }
1227
1228                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1229
1230                 if ( count( $this->mode ) === 0 ) {
1231                         // This should not happen.
1232                         throw new RuntimeException( 'Error extracting XMP, '
1233                                 . "encountered <$elm> with no mode" );
1234                 }
1235
1236                 switch ( $this->mode[0] ) {
1237                         case self::MODE_IGNORE:
1238                                 $this->startElementModeIgnore( $elm );
1239                                 break;
1240                         case self::MODE_SIMPLE:
1241                                 $this->startElementModeSimple( $elm, $attribs );
1242                                 break;
1243                         case self::MODE_INITIAL:
1244                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1245                                 break;
1246                         case self::MODE_STRUCT:
1247                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1248                                 break;
1249                         case self::MODE_BAG:
1250                         case self::MODE_BAGSTRUCT:
1251                                 $this->startElementModeBag( $elm );
1252                                 break;
1253                         case self::MODE_SEQ:
1254                                 $this->startElementModeSeq( $elm );
1255                                 break;
1256                         case self::MODE_LANG:
1257                                 $this->startElementModeLang( $elm );
1258                                 break;
1259                         case self::MODE_LI_LANG:
1260                                 $this->startElementModeLiLang( $elm, $attribs );
1261                                 break;
1262                         case self::MODE_LI:
1263                                 $this->startElementModeLi( $elm, $attribs );
1264                                 break;
1265                         case self::MODE_QDESC:
1266                                 $this->startElementModeQDesc( $elm );
1267                                 break;
1268                         default:
1269                                 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode[0] );
1270                 }
1271         }
1272
1273         // @codingStandardsIgnoreStart Generic.Files.LineLength
1274         /**
1275          * Process attributes.
1276          * Simple values can be stored as either a tag or attribute
1277          *
1278          * Often the initial "<rdf:Description>" tag just has all the simple
1279          * properties as attributes.
1280          *
1281          * @par Example:
1282          * @code
1283          * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1284          * @endcode
1285          *
1286          * @param array $attribs Array attribute=>value
1287          * @throws RuntimeException
1288          */
1289         // @codingStandardsIgnoreEnd
1290         private function doAttribs( $attribs ) {
1291                 // first check for rdf:parseType attribute, as that can change
1292                 // how the attributes are interperted.
1293
1294                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1295                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1296                         && $this->mode[0] === self::MODE_SIMPLE
1297                 ) {
1298                         // this is equivalent to having an inner rdf:Description
1299                         $this->mode[0] = self::MODE_QDESC;
1300                 }
1301                 foreach ( $attribs as $name => $val ) {
1302                         if ( strpos( $name, ' ' ) === false ) {
1303                                 // This shouldn't happen, but so far some old software forgets namespace
1304                                 // on rdf:about.
1305                                 $this->logger->info( __METHOD__ . ' Encountered non-namespaced attribute: '
1306                                         . " $name=\"$val\". Skipping. " );
1307                                 continue;
1308                         }
1309                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1310                         if ( $ns === self::NS_RDF ) {
1311                                 if ( $tag === 'value' || $tag === 'resource' ) {
1312                                         // resource is for url.
1313                                         // value attribute is a weird way of just putting the contents.
1314                                         $this->char( $this->xmlParser, $val );
1315                                 }
1316                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1317                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1318                                         throw new RuntimeException( __METHOD__
1319                                                 . " $ns:$tag found as attribute where not allowed" );
1320                                 }
1321                                 $this->saveValue( $ns, $tag, $val );
1322                         } else {
1323                                 $this->logger->debug( __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1324                         }
1325                 }
1326         }
1327
1328         /**
1329          * Given an extracted value, save it to results array
1330          *
1331          * note also uses $this->ancestorStruct and
1332          * $this->processingArray to determine what name to
1333          * save the value under. (in addition to $tag).
1334          *
1335          * @param string $ns Namespace of tag this is for
1336          * @param string $tag Tag name
1337          * @param string $val Value to save
1338          */
1339         private function saveValue( $ns, $tag, $val ) {
1340
1341                 $info =& $this->items[$ns][$tag];
1342                 $finalName = isset( $info['map_name'] )
1343                         ? $info['map_name'] : $tag;
1344                 if ( isset( $info['validate'] ) ) {
1345                         if ( is_array( $info['validate'] ) ) {
1346                                 $validate = $info['validate'];
1347                         } else {
1348                                 $validator = new XMPValidate( $this->logger );
1349                                 $validate = [ $validator, $info['validate'] ];
1350                         }
1351
1352                         if ( is_callable( $validate ) ) {
1353                                 call_user_func_array( $validate, [ $info, &$val, true ] );
1354                                 // the reasoning behind using &$val instead of using the return value
1355                                 // is to be consistent between here and validating structures.
1356                                 if ( is_null( $val ) ) {
1357                                         $this->logger->info( __METHOD__ . " <$ns:$tag> failed validation." );
1358
1359                                         return;
1360                                 }
1361                         } else {
1362                                 $this->logger->warning( __METHOD__ . " Validation function for $finalName ("
1363                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1364                         }
1365                 }
1366
1367                 if ( $this->ancestorStruct && $this->processingArray ) {
1368                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1369                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1370                 } elseif ( $this->ancestorStruct ) {
1371                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1372                 } elseif ( $this->processingArray ) {
1373                         if ( $this->itemLang === false ) {
1374                                 // normal array
1375                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1376                         } else {
1377                                 // lang array.
1378                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1379                         }
1380                 } else {
1381                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1382                 }
1383         }
1384 }