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 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 = [];
  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 = [];
  68
  69         /** @var array Array to hold results */
  70         private $results = [];
  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          * @var string
 134          */
 135         private $filename;
 136
 137         /**
 138          * Primary job is to initialize the XMLParser
 139          * @param LoggerInterface|null $logger
 140          * @param string $filename
 141          */
 142         function __construct( LoggerInterface $logger = null, $filename = 'unknown' ) {
 143                 if ( !function_exists( 'xml_parser_create_ns' ) ) {
 144                         // this should already be checked by this point
 145                         throw new RuntimeException( 'XMP support requires XML Parser' );
 146                 }
 147                 if ( $logger ) {
 148                         $this->setLogger( $logger );
 149                 } else {
 150                         $this->setLogger( new NullLogger() );
 151                 }
 152                 $this->filename = $filename;
 153
 154                 $this->items = XMPInfo::getItems();
 155
 156                 $this->resetXMLParser();
 157         }
 158
 159         public function setLogger( LoggerInterface $logger ) {
 160                 $this->logger = $logger;
 161         }
 162
 163         /**
 164          * free the XML parser.
 165          *
 166          * @note It is unclear to me if we really need to do this ourselves
 167          *  or if php garbage collection will automatically free the xmlParser
 168          *  when it is no longer needed.
 169          */
 170         private function destroyXMLParser() {
 171                 if ( $this->xmlParser ) {
 172                         xml_parser_free( $this->xmlParser );
 173                         $this->xmlParser = null;
 174                 }
 175         }
 176
 177         /**
 178          * Main use is if a single item has multiple xmp documents describing it.
 179          * For example in jpeg's with extendedXMP
 180          */
 181         private function resetXMLParser() {
 182                 $this->destroyXMLParser();
 183
 184                 $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' );
 185                 xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 );
 186                 xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 );
 187
 188                 xml_set_element_handler( $this->xmlParser,
 189                         [ $this, 'startElement' ],
 190                         [ $this, 'endElement' ] );
 191
 192                 xml_set_character_data_handler( $this->xmlParser, [ $this, 'char' ] );
 193
 194                 $this->parsable = self::PARSABLE_UNKNOWN;
 195                 $this->xmlParsableBuffer = '';
 196         }
 197
 198         /**
 199          * Check if this instance supports using this class
 200          * @return bool
 201          */
 202         public static function isSupported() {
 203                 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
 204         }
 205
 206         /** Get the result array. Do some post-processing before returning
 207          * the array, and transform any metadata that is special-cased.
 208          *
 209          * @return array Array of results as an array of arrays suitable for
 210          *    FormatMetadata::getFormattedData().
 211          */
 212         public function getResults() {
 213                 // xmp-special is for metadata that affects how stuff
 214                 // is extracted. For example xmpNote:HasExtendedXMP.
 215
 216                 // It is also used to handle photoshop:AuthorsPosition
 217                 // which is weird and really part of another property,
 218                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 219                 // The location fields also use it.
 220
 221                 $data = $this->results;
 222
 223                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 224                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 225                         && isset( $data['xmp-general']['Artist'][0] )
 226                 ) {
 227                         // Note, if there is more than one creator,
 228                         // this only applies to first. This also will
 229                         // only apply to the dc:Creator prop, not the
 230                         // exif:Artist prop.
 231
 232                         $data['xmp-general']['Artist'][0] =
 233                                 $data['xmp-special']['AuthorsPosition'] . ', '
 234                                 . $data['xmp-general']['Artist'][0];
 235                 }
 236
 237                 // Go through the LocationShown and LocationCreated
 238                 // changing it to the non-hierarchal form used by
 239                 // the other location fields.
 240
 241                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 242                         && is_array( $data['xmp-special']['LocationShown'][0] )
 243                 ) {
 244                         // the is_array is just paranoia. It should always
 245                         // be an array.
 246                         foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
 247                                 if ( !is_array( $loc ) ) {
 248                                         // To avoid copying over the _type meta-fields.
 249                                         continue;
 250                                 }
 251                                 foreach ( $loc as $field => $val ) {
 252                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 253                                 }
 254                         }
 255                 }
 256                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 257                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 258                 ) {
 259                         // the is_array is just paranoia. It should always
 260                         // be an array.
 261                         foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
 262                                 if ( !is_array( $loc ) ) {
 263                                         // To avoid copying over the _type meta-fields.
 264                                         continue;
 265                                 }
 266                                 foreach ( $loc as $field => $val ) {
 267                                         $data['xmp-general'][$field . 'Created'][] = $val;
 268                                 }
 269                         }
 270                 }
 271
 272                 // We don't want to return the special values, since they're
 273                 // special and not info to be stored about the file.
 274                 unset( $data['xmp-special'] );
 275
 276                 // Convert GPSAltitude to negative if below sea level.
 277                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
 278                         && isset( $data['xmp-exif']['GPSAltitude'] )
 279                 ) {
 280                         // Must convert to a real before multiplying by -1
 281                         // XMPValidate guarantees there will always be a '/' in this value.
 282                         list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
 283                         $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
 284
 285                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
 286                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 287                         }
 288                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 289                 }
 290
 291                 return $data;
 292         }
 293
 294         /**
 295          * Main function to call to parse XMP. Use getResults to
 296          * get results.
 297          *
 298          * Also catches any errors during processing, writes them to
 299          * debug log, blanks result array and returns false.
 300          *
 301          * @param string $content XMP data
 302          * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
 303          * @throws RuntimeException
 304          * @return bool Success.
 305          */
 306         public function parse( $content, $allOfIt = true ) {
 307                 if ( !$this->xmlParser ) {
 308                         $this->resetXMLParser();
 309                 }
 310                 try {
 311
 312                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 313                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 314                         if ( !$this->charset ) {
 315                                 $bom = [];
 316                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 317                                         $content, $bom )
 318                                 ) {
 319                                         switch ( $bom[0] ) {
 320                                                 case "\xFE\xFF":
 321                                                         $this->charset = 'UTF-16BE';
 322                                                         break;
 323                                                 case "\xFF\xFE":
 324                                                         $this->charset = 'UTF-16LE';
 325                                                         break;
 326                                                 case "\x00\x00\xFE\xFF":
 327                                                         $this->charset = 'UTF-32BE';
 328                                                         break;
 329                                                 case "\xFF\xFE\x00\x00":
 330                                                         $this->charset = 'UTF-32LE';
 331                                                         break;
 332                                                 case "\xEF\xBB\xBF":
 333                                                         $this->charset = 'UTF-8';
 334                                                         break;
 335                                                 default:
 336                                                         // this should be impossible to get to
 337                                                         throw new RuntimeException( "Invalid BOM" );
 338                                         }
 339                                 } else {
 340                                         // standard specifically says, if no bom assume utf-8
 341                                         $this->charset = 'UTF-8';
 342                                 }
 343                         }
 344                         if ( $this->charset !== 'UTF-8' ) {
 345                                 // don't convert if already utf-8
 346                                 Wikimedia\suppressWarnings();
 347                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 348                                 Wikimedia\restoreWarnings();
 349                         }
 350
 351                         // Ensure the XMP block does not have an xml doctype declaration, which
 352                         // could declare entities unsafe to parse with xml_parse (T85848/T71210).
 353                         if ( $this->parsable !== self::PARSABLE_OK ) {
 354                                 if ( $this->parsable === self::PARSABLE_NO ) {
 355                                         throw new RuntimeException( 'Unsafe doctype declaration in XML.' );
 356                                 }
 357
 358                                 $content = $this->xmlParsableBuffer . $content;
 359                                 if ( !$this->checkParseSafety( $content ) ) {
 360                                         if ( !$allOfIt && $this->parsable !== self::PARSABLE_NO ) {
 361                                                 // parse wasn't Unsuccessful yet, so return true
 362                                                 // in this case.
 363                                                 return true;
 364                                         }
 365                                         $msg = ( $this->parsable === self::PARSABLE_NO ) ?
 366                                                 'Unsafe doctype declaration in XML.' :
 367                                                 'No root element found in XML.';
 368                                         throw new RuntimeException( $msg );
 369                                 }
 370                         }
 371
 372                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 373                         if ( !$ok ) {
 374                                 $code = xml_get_error_code( $this->xmlParser );
 375                                 $error = xml_error_string( $code );
 376                                 $line = xml_get_current_line_number( $this->xmlParser );
 377                                 $col = xml_get_current_column_number( $this->xmlParser );
 378                                 $offset = xml_get_current_byte_index( $this->xmlParser );
 379
 380                                 $this->logger->info(
 381                                         '{method} : Error reading XMP content: {error} ' .
 382                                         '(file: {file}, line: {line} column: {column} ' .
 383                                         'byte offset: {offset})',
 384                                         [
 385                                                 'method' => __METHOD__,
 386                                                 'error_code' => $code,
 387                                                 'error' => $error,
 388                                                 'file' => $this->filename,
 389                                                 'line' => $line,
 390                                                 'column' => $col,
 391                                                 'offset' => $offset,
 392                                                 'content' => $content,
 393                                 ] );
 394                                 $this->results = []; // blank if error.
 395                                 $this->destroyXMLParser();
 396                                 return false;
 397                         }
 398                 } catch ( Exception $e ) {
 399                         $this->logger->warning(
 400                                 '{method} {exception}',
 401                                 [
 402                                         'method' => __METHOD__,
 403                                         'exception' => $e,
 404                                         'file' => $this->filename,
 405                                         'content' => $content,
 406                                 ]
 407                         );
 408                         $this->results = [];
 409                         return false;
 410                 }
 411                 if ( $allOfIt ) {
 412                         $this->destroyXMLParser();
 413                 }
 414
 415                 return true;
 416         }
 417
 418         /** Entry point for XMPExtended blocks in jpeg files
 419          *
 420          * @todo In serious need of testing
 421          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 422          * @param string $content XMPExtended block minus the namespace signature
 423          * @return bool If it succeeded.
 424          */
 425         public function parseExtended( $content ) {
 426                 // @todo FIXME: This is untested. Hard to find example files
 427                 // or programs that make such files..
 428                 $guid = substr( $content, 0, 32 );
 429                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 430                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid
 431                 ) {
 432                         $this->logger->info( __METHOD__ .
 433                                 " Ignoring XMPExtended block due to wrong guid (guid= '{guid}')",
 434                                         [
 435                                                 'guid' => $guid,
 436                                                 'file' => $this->filename,
 437                                         ]
 438                         );
 439
 440                         return false;
 441                 }
 442                 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 443
 444                 if ( !$len ||
 445                         $len['length'] < 4 ||
 446                         $len['offset'] < 0 ||
 447                         $len['offset'] > $len['length']
 448                 ) {
 449                         $this->logger->info(
 450                                 __METHOD__ . 'Error reading extended XMP block, invalid length or offset.',
 451                                 [ 'file' => $this->filename ]
 452                         );
 453
 454                         return false;
 455                 }
 456
 457                 // we're not very robust here. we should accept it in the wrong order.
 458                 // To quote the XMP standard:
 459                 // "A JPEG writer should write the ExtendedXMP marker segments in order,
 460                 // immediately following the StandardXMP. However, the JPEG standard
 461                 // does not require preservation of marker segment order. A robust JPEG
 462                 // reader should tolerate the marker segments in any order."
 463                 // On the other hand, the probability that an image will have more than
 464                 // 128k of metadata is rather low... so the probability that it will have
 465                 // > 128k, and be in the wrong order is very low...
 466
 467                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 468                         $this->logger->info( __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 469                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')',
 470                                 [ 'file' => $this->filename ]
 471                         );
 472
 473                         return false;
 474                 }
 475
 476                 if ( $len['offset'] === 0 ) {
 477                         // if we're starting the extended block, we've probably already
 478                         // done the XMPStandard block, so reset.
 479                         $this->resetXMLParser();
 480                 }
 481
 482                 $this->extendedXMPOffset += $len['length'];
 483
 484                 $actualContent = substr( $content, 40 );
 485
 486                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 487                         $atEnd = true;
 488                 } else {
 489                         $atEnd = false;
 490                 }
 491
 492                 $this->logger->debug(
 493                         __METHOD__ . 'Parsing a XMPExtended block',
 494                         [ 'file' => $this->filename ]
 495                 );
 496
 497                 return $this->parse( $actualContent, $atEnd );
 498         }
 499
 500         /**
 501          * Character data handler
 502          * Called whenever character data is found in the xmp document.
 503          *
 504          * does nothing if we're in MODE_IGNORE or if the data is whitespace
 505          * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 506          * data in the other modes).
 507          *
 508          * As an example, this happens when we encounter XMP like:
 509          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 510          * and are processing the 0/10 bit.
 511          *
 512          * @param resource $parser XMLParser reference to the xml parser
 513          * @param string $data Character data
 514          * @throws RuntimeException On invalid data
 515          */
 516         function char( $parser, $data ) {
 517                 $data = trim( $data );
 518                 if ( trim( $data ) === "" ) {
 519                         return;
 520                 }
 521
 522                 if ( !isset( $this->mode[0] ) ) {
 523                         throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
 524                 }
 525
 526                 if ( $this->mode[0] === self::MODE_IGNORE ) {
 527                         return;
 528                 }
 529
 530                 if ( $this->mode[0] !== self::MODE_SIMPLE
 531                         && $this->mode[0] !== self::MODE_QDESC
 532                 ) {
 533                         throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 534                 }
 535
 536                 // to check, how does this handle w.s.
 537                 if ( $this->charContent === false ) {
 538                         $this->charContent = $data;
 539                 } else {
 540                         $this->charContent .= $data;
 541                 }
 542         }
 543
 544         /**
 545          * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
 546          * contain a doctype declaration which could contain a dos attack if we
 547          * parse it and expand internal entities (T85848).
 548          *
 549          * @param string $content xml string to check for parse safety
 550          * @return bool true if the xml is safe to parse, false otherwise
 551          */
 552         private function checkParseSafety( $content ) {
 553                 $reader = new XMLReader();
 554                 $result = null;
 555
 556                 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
 557                 // instead of using XML().
 558                 $reader->open(
 559                         'data://text/plain,' . urlencode( $content ),
 560                         null,
 561                         LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
 562                 );
 563
 564                 $oldDisable = libxml_disable_entity_loader( true );
 565                 /** @noinspection PhpUnusedLocalVariableInspection */
 566                 $reset = new ScopedCallback(
 567                         'libxml_disable_entity_loader',
 568                         [ $oldDisable ]
 569                 );
 570                 $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false );
 571
 572                 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
 573                 // when parsing truncated XML, which causes unit tests to fail.
 574                 Wikimedia\suppressWarnings();
 575                 while ( $reader->read() ) {
 576                         if ( $reader->nodeType === XMLReader::ELEMENT ) {
 577                                 // Reached the first element without hitting a doctype declaration
 578                                 $this->parsable = self::PARSABLE_OK;
 579                                 $result = true;
 580                                 break;
 581                         }
 582                         if ( $reader->nodeType === XMLReader::DOC_TYPE ) {
 583                                 $this->parsable = self::PARSABLE_NO;
 584                                 $result = false;
 585                                 break;
 586                         }
 587                 }
 588                 Wikimedia\restoreWarnings();
 589
 590                 if ( !is_null( $result ) ) {
 591                         return $result;
 592                 }
 593
 594                 // Reached the end of the parsable xml without finding an element
 595                 // or doctype. Buffer and try again.
 596                 $this->parsable = self::PARSABLE_BUFFERING;
 597                 $this->xmlParsableBuffer = $content;
 598                 return false;
 599         }
 600
 601         /** When we hit a closing element in MODE_IGNORE
 602          * Check to see if this is the element we started to ignore,
 603          * in which case we get out of MODE_IGNORE
 604          *
 605          * @param string $elm Namespace of element followed by a space and then tag name of element.
 606          */
 607         private function endElementModeIgnore( $elm ) {
 608                 if ( $this->curItem[0] === $elm ) {
 609                         array_shift( $this->curItem );
 610                         array_shift( $this->mode );
 611                 }
 612         }
 613
 614         /**
 615          * Hit a closing element when in MODE_SIMPLE.
 616          * This generally means that we finished processing a
 617          * property value, and now have to save the result to the
 618          * results array
 619          *
 620          * For example, when processing:
 621          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 622          * this deals with when we hit </exif:DigitalZoomRatio>.
 623          *
 624          * Or it could be if we hit the end element of a property
 625          * of a compound data structure (like a member of an array).
 626          *
 627          * @param string $elm Namespace, space, and tag name.
 628          */
 629         private function endElementModeSimple( $elm ) {
 630                 if ( $this->charContent !== false ) {
 631                         if ( $this->processingArray ) {
 632                                 // if we're processing an array, use the original element
 633                                 // name instead of rdf:li.
 634                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 635                         } else {
 636                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 637                         }
 638                         $this->saveValue( $ns, $tag, $this->charContent );
 639
 640                         $this->charContent = false; // reset
 641                 }
 642                 array_shift( $this->curItem );
 643                 array_shift( $this->mode );
 644         }
 645
 646         /**
 647          * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 648          * generally means we've finished processing a nested structure.
 649          * resets some internal variables to indicate that.
 650          *
 651          * Note this means we hit the closing element not the "</rdf:Seq>".
 652          *
 653          * @par For example, when processing:
 654          * @code{,xml}
 655          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 656          *   </rdf:Seq> </exif:ISOSpeedRatings>
 657          * @endcode
 658          *
 659          * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
 660          *
 661          * @param string $elm Namespace . space . tag name.
 662          * @throws RuntimeException
 663          */
 664         private function endElementNested( $elm ) {
 665                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 666                  * in which case it could also be rdf:Description */
 667                 if ( $this->curItem[0] !== $elm
 668                         && !( $elm === self::NS_RDF . ' Description'
 669                                 && $this->mode[0] === self::MODE_STRUCT )
 670                 ) {
 671                         throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
 672                                 $this->curItem[0] . '>' );
 673                 }
 674
 675                 // Validate structures.
 676                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 677                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 678                         $info =& $this->items[$ns][$tag];
 679                         $finalName = $info['map_name'] ?? $tag;
 680
 681                         if ( is_array( $info['validate'] ) ) {
 682                                 $validate = $info['validate'];
 683                         } else {
 684                                 $validator = new XMPValidate( $this->logger );
 685                                 $validate = [ $validator, $info['validate'] ];
 686                         }
 687
 688                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 689                                 // This can happen if all the members of the struct failed validation.
 690                                 $this->logger->debug(
 691                                         __METHOD__ . " <$ns:$tag> has no valid members.",
 692                                         [ 'file' => $this->filename ]
 693                                 );
 694                         } elseif ( is_callable( $validate ) ) {
 695                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 696                                 call_user_func_array( $validate, [ $info, &$val, false ] );
 697                                 if ( is_null( $val ) ) {
 698                                         // the idea being the validation function will unset the variable if
 699                                         // its invalid.
 700                                         $this->logger->info(
 701                                                 __METHOD__ . " <$ns:$tag> failed validation.",
 702                                                 [ 'file' => $this->filename ]
 703                                         );
 704                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 705                                 }
 706                         } else {
 707                                 $this->logger->warning(
 708                                         __METHOD__ . " Validation function for $finalName (" .
 709                                         $validate[0] . '::' . $validate[1] . '()) is not callable.',
 710                                         [ 'file' => $this->filename ]
 711                                 );
 712                         }
 713                 }
 714
 715                 array_shift( $this->curItem );
 716                 array_shift( $this->mode );
 717                 $this->ancestorStruct = false;
 718                 $this->processingArray = false;
 719                 $this->itemLang = false;
 720         }
 721
 722         /**
 723          * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 724          * Add information about what type of element this is.
 725          *
 726          * Note we still have to hit the outer "</property>"
 727          *
 728          * @par For example, when processing:
 729          * @code{,xml}
 730          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 731          *   </rdf:Seq> </exif:ISOSpeedRatings>
 732          * @endcode
 733          *
 734          * This method is called when we hit the "</rdf:Seq>".
 735          * (For comparison, we call endElementModeSimple when we
 736          * hit the "</rdf:li>")
 737          *
 738          * @param string $elm Namespace . ' ' . element name
 739          * @throws RuntimeException
 740          */
 741         private function endElementModeLi( $elm ) {
 742                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 743                 $info = $this->items[$ns][$tag];
 744                 $finalName = $info['map_name'] ?? $tag;
 745
 746                 array_shift( $this->mode );
 747
 748                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 749                         $this->logger->debug(
 750                                 __METHOD__ . " Empty compund element $finalName.",
 751                                 [ 'file' => $this->filename ]
 752                         );
 753
 754                         return;
 755                 }
 756
 757                 if ( $elm === self::NS_RDF . ' Seq' ) {
 758                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 759                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 760                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 761                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 762                         // extra if needed as you could theoretically have a non-language alt.
 763                         if ( $info['mode'] === self::MODE_LANG ) {
 764                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 765                         }
 766                 } else {
 767                         throw new RuntimeException(
 768                                 __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm."
 769                         );
 770                 }
 771         }
 772
 773         /**
 774          * End element while in MODE_QDESC
 775          * mostly when ending an element when we have a simple value
 776          * that has qualifiers.
 777          *
 778          * Qualifiers aren't all that common, and we don't do anything
 779          * with them.
 780          *
 781          * @param string $elm Namespace and element
 782          */
 783         private function endElementModeQDesc( $elm ) {
 784                 if ( $elm === self::NS_RDF . ' value' ) {
 785                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 786                         $this->saveValue( $ns, $tag, $this->charContent );
 787
 788                         return;
 789                 } else {
 790                         array_shift( $this->mode );
 791                         array_shift( $this->curItem );
 792                 }
 793         }
 794
 795         /**
 796          * Handler for hitting a closing element.
 797          *
 798          * generally just calls a helper function depending on what
 799          * mode we're in.
 800          *
 801          * Ignores the outer wrapping elements that are optional in
 802          * xmp and have no meaning.
 803          *
 804          * @param resource $parser
 805          * @param string $elm Namespace . ' ' . element name
 806          * @throws RuntimeException
 807          */
 808         function endElement( $parser, $elm ) {
 809                 if ( $elm === ( self::NS_RDF . ' RDF' )
 810                         || $elm === 'adobe:ns:meta/ xmpmeta'
 811                         || $elm === 'adobe:ns:meta/ xapmeta'
 812                 ) {
 813                         // ignore these.
 814                         return;
 815                 }
 816
 817                 if ( $elm === self::NS_RDF . ' type' ) {
 818                         // these aren't really supported properly yet.
 819                         // However, it appears they almost never used.
 820                         $this->logger->info(
 821                                 __METHOD__ . ' encountered <rdf:type>',
 822                                 [ 'file' => $this->filename ]
 823                         );
 824                 }
 825
 826                 if ( strpos( $elm, ' ' ) === false ) {
 827                         // This probably shouldn't happen.
 828                         // However, there is a bug in an adobe product
 829                         // that forgets the namespace on some things.
 830                         // (Luckily they are unimportant things).
 831                         $this->logger->info(
 832                                 __METHOD__ . " Encountered </$elm> which has no namespace. Skipping.",
 833                                 [ 'file' => $this->filename ]
 834                         );
 835
 836                         return;
 837                 }
 838
 839                 if ( count( $this->mode ) === 0 ) {
 840                         // This should never ever happen and means
 841                         // there is a pretty major bug in this class.
 842                         throw new RuntimeException( 'Encountered end element with no mode' );
 843                 }
 844
 845                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 846                         // just to be paranoid. Should always have a curItem, except for initially
 847                         // (aka during MODE_INITAL).
 848                         throw new RuntimeException( "Hit end element </$elm> but no curItem" );
 849                 }
 850
 851                 switch ( $this->mode[0] ) {
 852                         case self::MODE_IGNORE:
 853                                 $this->endElementModeIgnore( $elm );
 854                                 break;
 855                         case self::MODE_SIMPLE:
 856                                 $this->endElementModeSimple( $elm );
 857                                 break;
 858                         case self::MODE_STRUCT:
 859                         case self::MODE_SEQ:
 860                         case self::MODE_BAG:
 861                         case self::MODE_LANG:
 862                         case self::MODE_BAGSTRUCT:
 863                                 $this->endElementNested( $elm );
 864                                 break;
 865                         case self::MODE_INITIAL:
 866                                 if ( $elm === self::NS_RDF . ' Description' ) {
 867                                         array_shift( $this->mode );
 868                                 } else {
 869                                         throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
 870                                 }
 871                                 break;
 872                         case self::MODE_LI:
 873                         case self::MODE_LI_LANG:
 874                                 $this->endElementModeLi( $elm );
 875                                 break;
 876                         case self::MODE_QDESC:
 877                                 $this->endElementModeQDesc( $elm );
 878                                 break;
 879                         default:
 880                                 $this->logger->info(
 881                                         __METHOD__ . " no mode (elm = $elm)",
 882                                         [ 'file' => $this->filename ]
 883                                 );
 884                                 break;
 885                 }
 886         }
 887
 888         /**
 889          * Hit an opening element while in MODE_IGNORE
 890          *
 891          * XMP is extensible, so ignore any tag we don't understand.
 892          *
 893          * Mostly ignores, unless we encounter the element that we are ignoring.
 894          * in which case we add it to the item stack, so we can ignore things
 895          * that are nested, correctly.
 896          *
 897          * @param string $elm Namespace . ' ' . tag name
 898          */
 899         private function startElementModeIgnore( $elm ) {
 900                 if ( $elm === $this->curItem[0] ) {
 901                         array_unshift( $this->curItem, $elm );
 902                         array_unshift( $this->mode, self::MODE_IGNORE );
 903                 }
 904         }
 905
 906         /**
 907          *  Start element in MODE_BAG (unordered array)
 908          * this should always be <rdf:Bag>
 909          *
 910          * @param string $elm Namespace . ' ' . tag
 911          * @throws RuntimeException If we have an element that's not <rdf:Bag>
 912          */
 913         private function startElementModeBag( $elm ) {
 914                 if ( $elm === self::NS_RDF . ' Bag' ) {
 915                         array_unshift( $this->mode, self::MODE_LI );
 916                 } else {
 917                         throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
 918                 }
 919         }
 920
 921         /**
 922          * Start element in MODE_SEQ (ordered array)
 923          * this should always be <rdf:Seq>
 924          *
 925          * @param string $elm Namespace . ' ' . tag
 926          * @throws RuntimeException If we have an element that's not <rdf:Seq>
 927          */
 928         private function startElementModeSeq( $elm ) {
 929                 if ( $elm === self::NS_RDF . ' Seq' ) {
 930                         array_unshift( $this->mode, self::MODE_LI );
 931                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 932                         # T29105
 933                         $this->logger->info(
 934                                 __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending' .
 935                                 ' it is a Seq, since some buggy software is known to screw this up.',
 936                                 [ 'file' => $this->filename ]
 937                         );
 938                         array_unshift( $this->mode, self::MODE_LI );
 939                 } else {
 940                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 941                 }
 942         }
 943
 944         /**
 945          * Start element in MODE_LANG (language alternative)
 946          * this should always be <rdf:Alt>
 947          *
 948          * This tag tends to be used for metadata like describe this
 949          * picture, which can be translated into multiple languages.
 950          *
 951          * XMP supports non-linguistic alternative selections,
 952          * which are really only used for thumbnails, which
 953          * we don't care about.
 954          *
 955          * @param string $elm Namespace . ' ' . tag
 956          * @throws RuntimeException If we have an element that's not <rdf:Alt>
 957          */
 958         private function startElementModeLang( $elm ) {
 959                 if ( $elm === self::NS_RDF . ' Alt' ) {
 960                         array_unshift( $this->mode, self::MODE_LI_LANG );
 961                 } else {
 962                         throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
 963                 }
 964         }
 965
 966         /**
 967          * Handle an opening element when in MODE_SIMPLE
 968          *
 969          * This should not happen often. This is for if a simple element
 970          * already opened has a child element. Could happen for a
 971          * qualified element.
 972          *
 973          * For example:
 974          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 975          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 976          *   </exif:DigitalZoomRatio>
 977          *
 978          * This method is called when processing the <rdf:Description> element
 979          *
 980          * @param string $elm Namespace and tag names separated by space.
 981          * @param array $attribs Attributes of the element.
 982          * @throws RuntimeException
 983          */
 984         private function startElementModeSimple( $elm, $attribs ) {
 985                 if ( $elm === self::NS_RDF . ' Description' ) {
 986                         // If this value has qualifiers
 987                         array_unshift( $this->mode, self::MODE_QDESC );
 988                         array_unshift( $this->curItem, $this->curItem[0] );
 989
 990                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 991                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 992                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 993                         }
 994                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 995                         // This should not be here.
 996                         throw new RuntimeException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 997                 } else {
 998                         // something else we don't recognize, like a qualifier maybe.
 999                         $this->logger->info( __METHOD__ .
1000                                 " Encountered element <{element}> where only expecting character data as value of {curitem}",
1001                                 [
1002                                         'element' => $elm,
1003                                         'curitem' => $this->curItem[0],
1004                                         'file' => $this->filename,
1005                                 ]
1006                         );
1007                         array_unshift( $this->mode, self::MODE_IGNORE );
1008                         array_unshift( $this->curItem, $elm );
1009                 }
1010         }
1011
1012         /**
1013          * Start an element when in MODE_QDESC.
1014          * This generally happens when a simple element has an inner
1015          * rdf:Description to hold qualifier elements.
1016          *
1017          * For example in:
1018          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
1019          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
1020          *   </exif:DigitalZoomRatio>
1021          * Called when processing the <rdf:value> or <foo:someQualifier>.
1022          *
1023          * @param string $elm Namespace and tag name separated by a space.
1024          */
1025         private function startElementModeQDesc( $elm ) {
1026                 if ( $elm === self::NS_RDF . ' value' ) {
1027                         return; // do nothing
1028                 } else {
1029                         // otherwise its a qualifier, which we ignore
1030                         array_unshift( $this->mode, self::MODE_IGNORE );
1031                         array_unshift( $this->curItem, $elm );
1032                 }
1033         }
1034
1035         /**
1036          * Starting an element when in MODE_INITIAL
1037          * This usually happens when we hit an element inside
1038          * the outer rdf:Description
1039          *
1040          * This is generally where most properties start.
1041          *
1042          * @param string $ns Namespace
1043          * @param string $tag Tag name (without namespace prefix)
1044          * @param array $attribs Array of attributes
1045          * @throws RuntimeException
1046          */
1047         private function startElementModeInitial( $ns, $tag, $attribs ) {
1048                 if ( $ns !== self::NS_RDF ) {
1049                         if ( isset( $this->items[$ns][$tag] ) ) {
1050                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
1051                                         // If this element is supposed to appear only as
1052                                         // a child of a structure, but appears here (not as
1053                                         // a child of a struct), then something weird is
1054                                         // happening, so ignore this element and its children.
1055
1056                                         $this->logger->info(
1057                                                 'Encountered <{element}> outside of its expected parent. Ignoring.',
1058                                                 [ 'element' => "$ns:$tag", 'file' => $this->filename ]
1059                                         );
1060
1061                                         array_unshift( $this->mode, self::MODE_IGNORE );
1062                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
1063
1064                                         return;
1065                                 }
1066                                 $mode = $this->items[$ns][$tag]['mode'];
1067                                 array_unshift( $this->mode, $mode );
1068                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1069                                 if ( $mode === self::MODE_STRUCT ) {
1070                                         $this->ancestorStruct = $this->items[$ns][$tag]['map_name'] ?? $tag;
1071                                 }
1072                                 if ( $this->charContent !== false ) {
1073                                         // Something weird.
1074                                         // Should not happen in valid XMP.
1075                                         throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1076                                 }
1077                         } else {
1078                                 // This element is not on our list of allowed elements so ignore.
1079                                 $this->logger->debug( __METHOD__ . ' Ignoring unrecognized element <{element}>.',
1080                                         [ 'element' => "$ns:$tag", 'file' => $this->filename ] );
1081                                 array_unshift( $this->mode, self::MODE_IGNORE );
1082                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1083
1084                                 return;
1085                         }
1086                 }
1087                 // process attributes
1088                 $this->doAttribs( $attribs );
1089         }
1090
1091         /**
1092          * Hit an opening element when in a Struct (MODE_STRUCT)
1093          * This is generally for fields of a compound property.
1094          *
1095          * Example of a struct (abbreviated; flash has more properties):
1096          *
1097          * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1098          *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1099          *
1100          * or:
1101          *
1102          * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1103          *  <exif:Mode>1</exif:Mode></exif:Flash>
1104          *
1105          * @param string $ns Namespace
1106          * @param string $tag Tag name (no ns)
1107          * @param array $attribs Array of attribs w/ values.
1108          * @throws RuntimeException
1109          */
1110         private function startElementModeStruct( $ns, $tag, $attribs ) {
1111                 if ( $ns !== self::NS_RDF ) {
1112                         if ( isset( $this->items[$ns][$tag] ) ) {
1113                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
1114                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] )
1115                                 ) {
1116                                         // This assumes that we don't have inter-namespace nesting
1117                                         // which we don't in all the properties we're interested in.
1118                                         throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1119                                                 . "> where it is not allowed." );
1120                                 }
1121                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
1122                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1123                                 if ( $this->charContent !== false ) {
1124                                         // Something weird.
1125                                         // Should not happen in valid XMP.
1126                                         throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1127                                                 $this->charContent . ")." );
1128                                 }
1129                         } else {
1130                                 array_unshift( $this->mode, self::MODE_IGNORE );
1131                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1132
1133                                 return;
1134                         }
1135                 }
1136
1137                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
1138                         $this->doAttribs( $attribs );
1139                         array_unshift( $this->mode, self::MODE_STRUCT );
1140                         array_unshift( $this->curItem, $this->curItem[0] );
1141                 }
1142         }
1143
1144         /**
1145          * opening element in MODE_LI
1146          * process elements of arrays.
1147          *
1148          * Example:
1149          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1150          *   </rdf:Seq> </exif:ISOSpeedRatings>
1151          * This method is called when we hit the <rdf:li> element.
1152          *
1153          * @param string $elm Namespace . ' ' . tagname
1154          * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1155          * @throws RuntimeException If gets a tag other than <rdf:li>
1156          */
1157         private function startElementModeLi( $elm, $attribs ) {
1158                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
1159                         throw new RuntimeException( "<rdf:li> expected but got $elm." );
1160                 }
1161
1162                 if ( !isset( $this->mode[1] ) ) {
1163                         // This should never ever ever happen. Checking for it
1164                         // to be paranoid.
1165                         throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1166                 }
1167
1168                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
1169                         // This list item contains a compound (STRUCT) value.
1170                         array_unshift( $this->mode, self::MODE_STRUCT );
1171                         array_unshift( $this->curItem, $elm );
1172                         $this->processingArray = true;
1173
1174                         if ( !isset( $this->curItem[1] ) ) {
1175                                 // be paranoid.
1176                                 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1177                         }
1178                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
1179                         $this->ancestorStruct = $this->items[$curNS][$curTag]['map_name'] ?? $curTag;
1180
1181                         $this->doAttribs( $attribs );
1182                 } else {
1183                         // Normal BAG or SEQ containing simple values.
1184                         array_unshift( $this->mode, self::MODE_SIMPLE );
1185                         // need to add curItem[0] on again since one is for the specific item
1186                         // and one is for the entire group.
1187                         array_unshift( $this->curItem, $this->curItem[0] );
1188                         $this->processingArray = true;
1189                 }
1190         }
1191
1192         /**
1193          * Opening element in MODE_LI_LANG.
1194          * process elements of language alternatives
1195          *
1196          * Example:
1197          * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1198          *  </rdf:li> </rdf:Alt> </dc:title>
1199          *
1200          * This method is called when we hit the <rdf:li> element.
1201          *
1202          * @param string $elm Namespace . ' ' . tag
1203          * @param array $attribs Array of elements (most importantly xml:lang)
1204          * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1205          */
1206         private function startElementModeLiLang( $elm, $attribs ) {
1207                 if ( $elm !== self::NS_RDF . ' li' ) {
1208                         throw new RuntimeException( __METHOD__ . " <rdf:li> expected but got $elm." );
1209                 }
1210                 if ( !isset( $attribs[self::NS_XML . ' lang'] )
1211                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] )
1212                 ) {
1213                         throw new RuntimeException( __METHOD__
1214                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1215                 }
1216
1217                 // Lang is case-insensitive.
1218                 $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] );
1219
1220                 // need to add curItem[0] on again since one is for the specific item
1221                 // and one is for the entire group.
1222                 array_unshift( $this->curItem, $this->curItem[0] );
1223                 array_unshift( $this->mode, self::MODE_SIMPLE );
1224                 $this->processingArray = true;
1225         }
1226
1227         /**
1228          * Hits an opening element.
1229          * Generally just calls a helper based on what MODE we're in.
1230          * Also does some initial set up for the wrapper element
1231          *
1232          * @param resource $parser
1233          * @param string $elm Namespace "<space>" element
1234          * @param array $attribs Attribute name => value
1235          * @throws RuntimeException
1236          */
1237         function startElement( $parser, $elm, $attribs ) {
1238                 if ( $elm === self::NS_RDF . ' RDF'
1239                         || $elm === 'adobe:ns:meta/ xmpmeta'
1240                         || $elm === 'adobe:ns:meta/ xapmeta'
1241                 ) {
1242                         /* ignore. */
1243                         return;
1244                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1245                         if ( count( $this->mode ) === 0 ) {
1246                                 // outer rdf:desc
1247                                 array_unshift( $this->mode, self::MODE_INITIAL );
1248                         }
1249                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1250                         // This doesn't support rdf:type properly.
1251                         // In practise I have yet to see a file that
1252                         // uses this element, however it is mentioned
1253                         // on page 25 of part 1 of the xmp standard.
1254                         // Also it seems as if exiv2 and exiftool do not support
1255                         // this either (That or I misunderstand the standard)
1256                         $this->logger->info(
1257                                 __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported',
1258                                 [ 'file' => $this->filename ]
1259                         );
1260                 }
1261
1262                 if ( strpos( $elm, ' ' ) === false ) {
1263                         // This probably shouldn't happen.
1264                         $this->logger->info(
1265                                 __METHOD__ . " Encountered <$elm> which has no namespace. Skipping.",
1266                                 [ 'file' => $this->filename ]
1267                         );
1268
1269                         return;
1270                 }
1271
1272                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1273
1274                 if ( count( $this->mode ) === 0 ) {
1275                         // This should not happen.
1276                         throw new RuntimeException( 'Error extracting XMP, '
1277                                 . "encountered <$elm> with no mode" );
1278                 }
1279
1280                 switch ( $this->mode[0] ) {
1281                         case self::MODE_IGNORE:
1282                                 $this->startElementModeIgnore( $elm );
1283                                 break;
1284                         case self::MODE_SIMPLE:
1285                                 $this->startElementModeSimple( $elm, $attribs );
1286                                 break;
1287                         case self::MODE_INITIAL:
1288                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1289                                 break;
1290                         case self::MODE_STRUCT:
1291                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1292                                 break;
1293                         case self::MODE_BAG:
1294                         case self::MODE_BAGSTRUCT:
1295                                 $this->startElementModeBag( $elm );
1296                                 break;
1297                         case self::MODE_SEQ:
1298                                 $this->startElementModeSeq( $elm );
1299                                 break;
1300                         case self::MODE_LANG:
1301                                 $this->startElementModeLang( $elm );
1302                                 break;
1303                         case self::MODE_LI_LANG:
1304                                 $this->startElementModeLiLang( $elm, $attribs );
1305                                 break;
1306                         case self::MODE_LI:
1307                                 $this->startElementModeLi( $elm, $attribs );
1308                                 break;
1309                         case self::MODE_QDESC:
1310                                 $this->startElementModeQDesc( $elm );
1311                                 break;
1312                         default:
1313                                 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode[0] );
1314                 }
1315         }
1316
1317         // phpcs:disable Generic.Files.LineLength
1318         /**
1319          * Process attributes.
1320          * Simple values can be stored as either a tag or attribute
1321          *
1322          * Often the initial "<rdf:Description>" tag just has all the simple
1323          * properties as attributes.
1324          *
1325          * @par Example:
1326          * @code
1327          * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1328          * @endcode
1329          *
1330          * @param array $attribs Array attribute=>value
1331          * @throws RuntimeException
1332          */
1333         // phpcs:enable
1334         private function doAttribs( $attribs ) {
1335                 // first check for rdf:parseType attribute, as that can change
1336                 // how the attributes are interperted.
1337
1338                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1339                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1340                         && $this->mode[0] === self::MODE_SIMPLE
1341                 ) {
1342                         // this is equivalent to having an inner rdf:Description
1343                         $this->mode[0] = self::MODE_QDESC;
1344                 }
1345                 foreach ( $attribs as $name => $val ) {
1346                         if ( strpos( $name, ' ' ) === false ) {
1347                                 // This shouldn't happen, but so far some old software forgets namespace
1348                                 // on rdf:about.
1349                                 $this->logger->info(
1350                                         __METHOD__ . ' Encountered non-namespaced attribute: ' .
1351                                         " $name=\"$val\". Skipping. ",
1352                                         [ 'file' => $this->filename ]
1353                                 );
1354                                 continue;
1355                         }
1356                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1357                         if ( $ns === self::NS_RDF ) {
1358                                 if ( $tag === 'value' || $tag === 'resource' ) {
1359                                         // resource is for url.
1360                                         // value attribute is a weird way of just putting the contents.
1361                                         $this->char( $this->xmlParser, $val );
1362                                 }
1363                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1364                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1365                                         throw new RuntimeException( __METHOD__
1366                                                 . " $ns:$tag found as attribute where not allowed" );
1367                                 }
1368                                 $this->saveValue( $ns, $tag, $val );
1369                         } else {
1370                                 $this->logger->debug(
1371                                         __METHOD__ . " Ignoring unrecognized element <$ns:$tag>.",
1372                                         [ 'file' => $this->filename ]
1373                                 );
1374                         }
1375                 }
1376         }
1377
1378         /**
1379          * Given an extracted value, save it to results array
1380          *
1381          * note also uses $this->ancestorStruct and
1382          * $this->processingArray to determine what name to
1383          * save the value under. (in addition to $tag).
1384          *
1385          * @param string $ns Namespace of tag this is for
1386          * @param string $tag Tag name
1387          * @param string $val Value to save
1388          */
1389         private function saveValue( $ns, $tag, $val ) {
1390                 $info =& $this->items[$ns][$tag];
1391                 $finalName = $info['map_name'] ?? $tag;
1392                 if ( isset( $info['validate'] ) ) {
1393                         if ( is_array( $info['validate'] ) ) {
1394                                 $validate = $info['validate'];
1395                         } else {
1396                                 $validator = new XMPValidate( $this->logger );
1397                                 $validate = [ $validator, $info['validate'] ];
1398                         }
1399
1400                         if ( is_callable( $validate ) ) {
1401                                 call_user_func_array( $validate, [ $info, &$val, true ] );
1402                                 // the reasoning behind using &$val instead of using the return value
1403                                 // is to be consistent between here and validating structures.
1404                                 if ( is_null( $val ) ) {
1405                                         $this->logger->info(
1406                                                 __METHOD__ . " <$ns:$tag> failed validation.",
1407                                                 [ 'file' => $this->filename ]
1408                                         );
1409
1410                                         return;
1411                                 }
1412                         } else {
1413                                 $this->logger->warning(
1414                                         __METHOD__ . " Validation function for $finalName (" .
1415                                         $validate[0] . '::' . $validate[1] . '()) is not callable.',
1416                                         [ 'file' => $this->filename ]
1417                                 );
1418                         }
1419                 }
1420
1421                 if ( $this->ancestorStruct && $this->processingArray ) {
1422                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1423                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1424                 } elseif ( $this->ancestorStruct ) {
1425                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1426                 } elseif ( $this->processingArray ) {
1427                         if ( $this->itemLang === false ) {
1428                                 // normal array
1429                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1430                         } else {
1431                                 // lang array.
1432                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1433                         }
1434                 } else {
1435                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1436                 }
1437         }
1438 }