3 * Reader for XMP data containing properties relevant to images.
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.
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.
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
24 use Psr\Log\LoggerAwareInterface
;
25 use Psr\Log\LoggerInterface
;
26 use Psr\Log\NullLogger
;
27 use Wikimedia\ScopedCallback
;
30 * Class for reading xmp data containing properties relevant to
31 * images, and spitting out an array that FormatMetadata accepts.
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
40 * The public methods one would call in this class are
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).
47 * Outputs a results array.
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
53 class XMPReader
implements LoggerAwareInterface
{
54 /** @var array XMP item configuration array */
57 /** @var array Array to hold the current element (and previous element, and so on) */
58 private $curItem = [];
60 /** @var bool|string The structure name when processing nested structures. */
61 private $ancestorStruct = false;
63 /** @var bool|string Temporary holder for character data that appears in xmp doc. */
64 private $charContent = false;
66 /** @var array Stores the state the xmpreader is in (see MODE_FOO constants) */
69 /** @var array Array to hold results */
70 private $results = [];
72 /** @var bool If we're doing a seq or bag. */
73 private $processingArray = false;
75 /** @var bool|string Used for lang alts only */
76 private $itemLang = false;
78 /** @var resource A resource handle for the XML parser */
81 /** @var bool|string Character set like 'UTF-8' */
82 private $charset = false;
85 private $extendedXMPOffset = 0;
87 /** @var int Flag determining if the XMP is safe to parse **/
88 private $parsable = 0;
90 /** @var string Buffer of XML to parse **/
91 private $xmlParsableBuffer = '';
94 * These are various mode constants.
95 * they are used to figure out what to do
96 * with an element when its encountered.
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.
102 const MODE_INITIAL
= 0;
103 const MODE_IGNORE
= 1;
105 const MODE_LI_LANG
= 3;
106 const MODE_QDESC
= 4;
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.
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';
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;
128 * @var LoggerInterface
133 * Primary job is to initialize the XMLParser
134 * @param LoggerInterface|null $logger
136 function __construct( LoggerInterface
$logger = null ) {
137 if ( !function_exists( 'xml_parser_create_ns' ) ) {
138 // this should already be checked by this point
139 throw new RuntimeException( 'XMP support requires XML Parser' );
142 $this->setLogger( $logger );
144 $this->setLogger( new NullLogger() );
147 $this->items
= XMPInfo
::getItems();
149 $this->resetXMLParser();
152 public function setLogger( LoggerInterface
$logger ) {
153 $this->logger
= $logger;
157 * free the XML parser.
159 * @note It is unclear to me if we really need to do this ourselves
160 * or if php garbage collection will automatically free the xmlParser
161 * when it is no longer needed.
163 private function destroyXMLParser() {
164 if ( $this->xmlParser
) {
165 xml_parser_free( $this->xmlParser
);
166 $this->xmlParser
= null;
171 * Main use is if a single item has multiple xmp documents describing it.
172 * For example in jpeg's with extendedXMP
174 private function resetXMLParser() {
175 $this->destroyXMLParser();
177 $this->xmlParser
= xml_parser_create_ns( 'UTF-8', ' ' );
178 xml_parser_set_option( $this->xmlParser
, XML_OPTION_CASE_FOLDING
, 0 );
179 xml_parser_set_option( $this->xmlParser
, XML_OPTION_SKIP_WHITE
, 1 );
181 xml_set_element_handler( $this->xmlParser
,
182 [ $this, 'startElement' ],
183 [ $this, 'endElement' ] );
185 xml_set_character_data_handler( $this->xmlParser
, [ $this, 'char' ] );
187 $this->parsable
= self
::PARSABLE_UNKNOWN
;
188 $this->xmlParsableBuffer
= '';
192 * Check if this instance supports using this class
195 public static function isSupported() {
196 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
199 /** Get the result array. Do some post-processing before returning
200 * the array, and transform any metadata that is special-cased.
202 * @return array Array of results as an array of arrays suitable for
203 * FormatMetadata::getFormattedData().
205 public function getResults() {
206 // xmp-special is for metadata that affects how stuff
207 // is extracted. For example xmpNote:HasExtendedXMP.
209 // It is also used to handle photoshop:AuthorsPosition
210 // which is weird and really part of another property,
211 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
212 // The location fields also use it.
214 $data = $this->results
;
216 if ( isset( $data['xmp-special']['AuthorsPosition'] )
217 && is_string( $data['xmp-special']['AuthorsPosition'] )
218 && isset( $data['xmp-general']['Artist'][0] )
220 // Note, if there is more than one creator,
221 // this only applies to first. This also will
222 // only apply to the dc:Creator prop, not the
225 $data['xmp-general']['Artist'][0] =
226 $data['xmp-special']['AuthorsPosition'] . ', '
227 . $data['xmp-general']['Artist'][0];
230 // Go through the LocationShown and LocationCreated
231 // changing it to the non-hierarchal form used by
232 // the other location fields.
234 if ( isset( $data['xmp-special']['LocationShown'][0] )
235 && is_array( $data['xmp-special']['LocationShown'][0] )
237 // the is_array is just paranoia. It should always
239 foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
240 if ( !is_array( $loc ) ) {
241 // To avoid copying over the _type meta-fields.
244 foreach ( $loc as $field => $val ) {
245 $data['xmp-general'][$field . 'Dest'][] = $val;
249 if ( isset( $data['xmp-special']['LocationCreated'][0] )
250 && is_array( $data['xmp-special']['LocationCreated'][0] )
252 // the is_array is just paranoia. It should always
254 foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
255 if ( !is_array( $loc ) ) {
256 // To avoid copying over the _type meta-fields.
259 foreach ( $loc as $field => $val ) {
260 $data['xmp-general'][$field . 'Created'][] = $val;
265 // We don't want to return the special values, since they're
266 // special and not info to be stored about the file.
267 unset( $data['xmp-special'] );
269 // Convert GPSAltitude to negative if below sea level.
270 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
271 && isset( $data['xmp-exif']['GPSAltitude'] )
273 // Must convert to a real before multiplying by -1
274 // XMPValidate guarantees there will always be a '/' in this value.
275 list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
276 $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
278 if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
279 $data['xmp-exif']['GPSAltitude'] *= -1;
281 unset( $data['xmp-exif']['GPSAltitudeRef'] );
288 * Main function to call to parse XMP. Use getResults to
291 * Also catches any errors during processing, writes them to
292 * debug log, blanks result array and returns false.
294 * @param string $content XMP data
295 * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
296 * @throws RuntimeException
297 * @return bool Success.
299 public function parse( $content, $allOfIt = true ) {
300 if ( !$this->xmlParser
) {
301 $this->resetXMLParser();
305 // detect encoding by looking for BOM which is supposed to be in processing instruction.
306 // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
307 if ( !$this->charset
) {
309 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
314 $this->charset
= 'UTF-16BE';
317 $this->charset
= 'UTF-16LE';
319 case "\x00\x00\xFE\xFF":
320 $this->charset
= 'UTF-32BE';
322 case "\xFF\xFE\x00\x00":
323 $this->charset
= 'UTF-32LE';
326 $this->charset
= 'UTF-8';
329 // this should be impossible to get to
330 throw new RuntimeException( "Invalid BOM" );
333 // standard specifically says, if no bom assume utf-8
334 $this->charset
= 'UTF-8';
337 if ( $this->charset
!== 'UTF-8' ) {
338 // don't convert if already utf-8
339 MediaWiki\
suppressWarnings();
340 $content = iconv( $this->charset
, 'UTF-8//IGNORE', $content );
341 MediaWiki\restoreWarnings
();
344 // Ensure the XMP block does not have an xml doctype declaration, which
345 // could declare entities unsafe to parse with xml_parse (T85848/T71210).
346 if ( $this->parsable
!== self
::PARSABLE_OK
) {
347 if ( $this->parsable
=== self
::PARSABLE_NO
) {
348 throw new RuntimeException( 'Unsafe doctype declaration in XML.' );
351 $content = $this->xmlParsableBuffer
. $content;
352 if ( !$this->checkParseSafety( $content ) ) {
353 if ( !$allOfIt && $this->parsable
!== self
::PARSABLE_NO
) {
354 // parse wasn't Unsuccessful yet, so return true
358 $msg = ( $this->parsable
=== self
::PARSABLE_NO
) ?
359 'Unsafe doctype declaration in XML.' :
360 'No root element found in XML.';
361 throw new RuntimeException( $msg );
365 $ok = xml_parse( $this->xmlParser
, $content, $allOfIt );
367 $code = xml_get_error_code( $this->xmlParser
);
368 $error = xml_error_string( $code );
369 $line = xml_get_current_line_number( $this->xmlParser
);
370 $col = xml_get_current_column_number( $this->xmlParser
);
371 $offset = xml_get_current_byte_index( $this->xmlParser
);
373 $this->logger
->warning(
374 '{method} : Error reading XMP content: {error} ' .
375 '(line: {line} column: {column} byte offset: {offset})',
377 'method' => __METHOD__
,
378 'error_code' => $code,
383 'content' => $content,
385 $this->results
= []; // blank if error.
386 $this->destroyXMLParser();
389 } catch ( Exception
$e ) {
390 $this->logger
->warning(
391 '{method} {exception}',
393 'method' => __METHOD__
,
395 'content' => $content,
402 $this->destroyXMLParser();
408 /** Entry point for XMPExtended blocks in jpeg files
410 * @todo In serious need of testing
411 * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
412 * @param string $content XMPExtended block minus the namespace signature
413 * @return bool If it succeeded.
415 public function parseExtended( $content ) {
416 // @todo FIXME: This is untested. Hard to find example files
417 // or programs that make such files..
418 $guid = substr( $content, 0, 32 );
419 if ( !isset( $this->results
['xmp-special']['HasExtendedXMP'] )
420 ||
$this->results
['xmp-special']['HasExtendedXMP'] !== $guid
422 $this->logger
->info( __METHOD__
.
423 " Ignoring XMPExtended block due to wrong guid (guid= '{guid}')",
424 [ 'guid' => 'guid' ] );
428 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
431 $len['length'] < 4 ||
432 $len['offset'] < 0 ||
433 $len['offset'] > $len['length']
436 __METHOD__
. 'Error reading extended XMP block, invalid length or offset.'
442 // we're not very robust here. we should accept it in the wrong order.
443 // To quote the XMP standard:
444 // "A JPEG writer should write the ExtendedXMP marker segments in order,
445 // immediately following the StandardXMP. However, the JPEG standard
446 // does not require preservation of marker segment order. A robust JPEG
447 // reader should tolerate the marker segments in any order."
448 // On the other hand, the probability that an image will have more than
449 // 128k of metadata is rather low... so the probability that it will have
450 // > 128k, and be in the wrong order is very low...
452 if ( $len['offset'] !== $this->extendedXMPOffset
) {
453 $this->logger
->info( __METHOD__
. 'Ignoring XMPExtended block due to wrong order. (Offset was '
454 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset
. ')' );
459 if ( $len['offset'] === 0 ) {
460 // if we're starting the extended block, we've probably already
461 // done the XMPStandard block, so reset.
462 $this->resetXMLParser();
465 $this->extendedXMPOffset +
= $len['length'];
467 $actualContent = substr( $content, 40 );
469 if ( $this->extendedXMPOffset
=== strlen( $actualContent ) ) {
475 $this->logger
->debug( __METHOD__
. 'Parsing a XMPExtended block' );
477 return $this->parse( $actualContent, $atEnd );
481 * Character data handler
482 * Called whenever character data is found in the xmp document.
484 * does nothing if we're in MODE_IGNORE or if the data is whitespace
485 * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
486 * data in the other modes).
488 * As an example, this happens when we encounter XMP like:
489 * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
490 * and are processing the 0/10 bit.
492 * @param resource $parser XMLParser reference to the xml parser
493 * @param string $data Character data
494 * @throws RuntimeException On invalid data
496 function char( $parser, $data ) {
497 $data = trim( $data );
498 if ( trim( $data ) === "" ) {
502 if ( !isset( $this->mode
[0] ) ) {
503 throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
506 if ( $this->mode
[0] === self
::MODE_IGNORE
) {
510 if ( $this->mode
[0] !== self
::MODE_SIMPLE
511 && $this->mode
[0] !== self
::MODE_QDESC
513 throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode
[0] . ')' );
516 // to check, how does this handle w.s.
517 if ( $this->charContent
=== false ) {
518 $this->charContent
= $data;
520 $this->charContent
.= $data;
525 * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
526 * contain a doctype declaration which could contain a dos attack if we
527 * parse it and expand internal entities (T85848).
529 * @param string $content xml string to check for parse safety
530 * @return bool true if the xml is safe to parse, false otherwise
532 private function checkParseSafety( $content ) {
533 $reader = new XMLReader();
536 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
537 // instead of using XML().
539 'data://text/plain,' . urlencode( $content ),
541 LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
544 $oldDisable = libxml_disable_entity_loader( true );
545 /** @noinspection PhpUnusedLocalVariableInspection */
546 $reset = new ScopedCallback(
547 'libxml_disable_entity_loader',
550 $reader->setParserProperty( XMLReader
::SUBST_ENTITIES
, false );
552 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
553 // when parsing truncated XML, which causes unit tests to fail.
554 MediaWiki\
suppressWarnings();
555 while ( $reader->read() ) {
556 if ( $reader->nodeType
=== XMLReader
::ELEMENT
) {
557 // Reached the first element without hitting a doctype declaration
558 $this->parsable
= self
::PARSABLE_OK
;
562 if ( $reader->nodeType
=== XMLReader
::DOC_TYPE
) {
563 $this->parsable
= self
::PARSABLE_NO
;
568 MediaWiki\restoreWarnings
();
570 if ( !is_null( $result ) ) {
574 // Reached the end of the parsable xml without finding an element
575 // or doctype. Buffer and try again.
576 $this->parsable
= self
::PARSABLE_BUFFERING
;
577 $this->xmlParsableBuffer
= $content;
581 /** When we hit a closing element in MODE_IGNORE
582 * Check to see if this is the element we started to ignore,
583 * in which case we get out of MODE_IGNORE
585 * @param string $elm Namespace of element followed by a space and then tag name of element.
587 private function endElementModeIgnore( $elm ) {
588 if ( $this->curItem
[0] === $elm ) {
589 array_shift( $this->curItem
);
590 array_shift( $this->mode
);
595 * Hit a closing element when in MODE_SIMPLE.
596 * This generally means that we finished processing a
597 * property value, and now have to save the result to the
600 * For example, when processing:
601 * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
602 * this deals with when we hit </exif:DigitalZoomRatio>.
604 * Or it could be if we hit the end element of a property
605 * of a compound data structure (like a member of an array).
607 * @param string $elm Namespace, space, and tag name.
609 private function endElementModeSimple( $elm ) {
610 if ( $this->charContent
!== false ) {
611 if ( $this->processingArray
) {
612 // if we're processing an array, use the original element
613 // name instead of rdf:li.
614 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
616 list( $ns, $tag ) = explode( ' ', $elm, 2 );
618 $this->saveValue( $ns, $tag, $this->charContent
);
620 $this->charContent
= false; // reset
622 array_shift( $this->curItem
);
623 array_shift( $this->mode
);
627 * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
628 * generally means we've finished processing a nested structure.
629 * resets some internal variables to indicate that.
631 * Note this means we hit the closing element not the "</rdf:Seq>".
633 * @par For example, when processing:
635 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
636 * </rdf:Seq> </exif:ISOSpeedRatings>
639 * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
641 * @param string $elm Namespace . space . tag name.
642 * @throws RuntimeException
644 private function endElementNested( $elm ) {
645 /* cur item must be the same as $elm, unless if in MODE_STRUCT
646 * in which case it could also be rdf:Description */
647 if ( $this->curItem
[0] !== $elm
648 && !( $elm === self
::NS_RDF
. ' Description'
649 && $this->mode
[0] === self
::MODE_STRUCT
)
651 throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
652 $this->curItem
[0] . '>' );
655 // Validate structures.
656 list( $ns, $tag ) = explode( ' ', $elm, 2 );
657 if ( isset( $this->items
[$ns][$tag]['validate'] ) ) {
658 $info =& $this->items
[$ns][$tag];
659 $finalName = isset( $info['map_name'] )
660 ?
$info['map_name'] : $tag;
662 if ( is_array( $info['validate'] ) ) {
663 $validate = $info['validate'];
665 $validator = new XMPValidate( $this->logger
);
666 $validate = [ $validator, $info['validate'] ];
669 if ( !isset( $this->results
['xmp-' . $info['map_group']][$finalName] ) ) {
670 // This can happen if all the members of the struct failed validation.
671 $this->logger
->debug( __METHOD__
. " <$ns:$tag> has no valid members." );
672 } elseif ( is_callable( $validate ) ) {
673 $val =& $this->results
['xmp-' . $info['map_group']][$finalName];
674 call_user_func_array( $validate, [ $info, &$val, false ] );
675 if ( is_null( $val ) ) {
676 // the idea being the validation function will unset the variable if
678 $this->logger
->info( __METHOD__
. " <$ns:$tag> failed validation." );
679 unset( $this->results
['xmp-' . $info['map_group']][$finalName] );
682 $this->logger
->warning( __METHOD__
. " Validation function for $finalName ("
683 . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
687 array_shift( $this->curItem
);
688 array_shift( $this->mode
);
689 $this->ancestorStruct
= false;
690 $this->processingArray
= false;
691 $this->itemLang
= false;
695 * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
696 * Add information about what type of element this is.
698 * Note we still have to hit the outer "</property>"
700 * @par For example, when processing:
702 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
703 * </rdf:Seq> </exif:ISOSpeedRatings>
706 * This method is called when we hit the "</rdf:Seq>".
707 * (For comparison, we call endElementModeSimple when we
708 * hit the "</rdf:li>")
710 * @param string $elm Namespace . ' ' . element name
711 * @throws RuntimeException
713 private function endElementModeLi( $elm ) {
714 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
715 $info = $this->items
[$ns][$tag];
716 $finalName = isset( $info['map_name'] )
717 ?
$info['map_name'] : $tag;
719 array_shift( $this->mode
);
721 if ( !isset( $this->results
['xmp-' . $info['map_group']][$finalName] ) ) {
722 $this->logger
->debug( __METHOD__
. " Empty compund element $finalName." );
727 if ( $elm === self
::NS_RDF
. ' Seq' ) {
728 $this->results
['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
729 } elseif ( $elm === self
::NS_RDF
. ' Bag' ) {
730 $this->results
['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
731 } elseif ( $elm === self
::NS_RDF
. ' Alt' ) {
732 // extra if needed as you could theoretically have a non-language alt.
733 if ( $info['mode'] === self
::MODE_LANG
) {
734 $this->results
['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
737 throw new RuntimeException(
738 __METHOD__
. " expected </rdf:seq> or </rdf:bag> but instead got $elm."
744 * End element while in MODE_QDESC
745 * mostly when ending an element when we have a simple value
746 * that has qualifiers.
748 * Qualifiers aren't all that common, and we don't do anything
751 * @param string $elm Namespace and element
753 private function endElementModeQDesc( $elm ) {
754 if ( $elm === self
::NS_RDF
. ' value' ) {
755 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
756 $this->saveValue( $ns, $tag, $this->charContent
);
760 array_shift( $this->mode
);
761 array_shift( $this->curItem
);
766 * Handler for hitting a closing element.
768 * generally just calls a helper function depending on what
771 * Ignores the outer wrapping elements that are optional in
772 * xmp and have no meaning.
774 * @param resource $parser
775 * @param string $elm Namespace . ' ' . element name
776 * @throws RuntimeException
778 function endElement( $parser, $elm ) {
779 if ( $elm === ( self
::NS_RDF
. ' RDF' )
780 ||
$elm === 'adobe:ns:meta/ xmpmeta'
781 ||
$elm === 'adobe:ns:meta/ xapmeta'
787 if ( $elm === self
::NS_RDF
. ' type' ) {
788 // these aren't really supported properly yet.
789 // However, it appears they almost never used.
790 $this->logger
->info( __METHOD__
. ' encountered <rdf:type>' );
793 if ( strpos( $elm, ' ' ) === false ) {
794 // This probably shouldn't happen.
795 // However, there is a bug in an adobe product
796 // that forgets the namespace on some things.
797 // (Luckily they are unimportant things).
798 $this->logger
->info( __METHOD__
. " Encountered </$elm> which has no namespace. Skipping." );
803 if ( count( $this->mode
[0] ) === 0 ) {
804 // This should never ever happen and means
805 // there is a pretty major bug in this class.
806 throw new RuntimeException( 'Encountered end element with no mode' );
809 if ( count( $this->curItem
) == 0 && $this->mode
[0] !== self
::MODE_INITIAL
) {
810 // just to be paranoid. Should always have a curItem, except for initially
811 // (aka during MODE_INITAL).
812 throw new RuntimeException( "Hit end element </$elm> but no curItem" );
815 switch ( $this->mode
[0] ) {
816 case self
::MODE_IGNORE
:
817 $this->endElementModeIgnore( $elm );
819 case self
::MODE_SIMPLE
:
820 $this->endElementModeSimple( $elm );
822 case self
::MODE_STRUCT
:
825 case self
::MODE_LANG
:
826 case self
::MODE_BAGSTRUCT
:
827 $this->endElementNested( $elm );
829 case self
::MODE_INITIAL
:
830 if ( $elm === self
::NS_RDF
. ' Description' ) {
831 array_shift( $this->mode
);
833 throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
837 case self
::MODE_LI_LANG
:
838 $this->endElementModeLi( $elm );
840 case self
::MODE_QDESC
:
841 $this->endElementModeQDesc( $elm );
844 $this->logger
->warning( __METHOD__
. " no mode (elm = $elm)" );
850 * Hit an opening element while in MODE_IGNORE
852 * XMP is extensible, so ignore any tag we don't understand.
854 * Mostly ignores, unless we encounter the element that we are ignoring.
855 * in which case we add it to the item stack, so we can ignore things
856 * that are nested, correctly.
858 * @param string $elm Namespace . ' ' . tag name
860 private function startElementModeIgnore( $elm ) {
861 if ( $elm === $this->curItem
[0] ) {
862 array_unshift( $this->curItem
, $elm );
863 array_unshift( $this->mode
, self
::MODE_IGNORE
);
868 * Start element in MODE_BAG (unordered array)
869 * this should always be <rdf:Bag>
871 * @param string $elm Namespace . ' ' . tag
872 * @throws RuntimeException If we have an element that's not <rdf:Bag>
874 private function startElementModeBag( $elm ) {
875 if ( $elm === self
::NS_RDF
. ' Bag' ) {
876 array_unshift( $this->mode
, self
::MODE_LI
);
878 throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
883 * Start element in MODE_SEQ (ordered array)
884 * this should always be <rdf:Seq>
886 * @param string $elm Namespace . ' ' . tag
887 * @throws RuntimeException If we have an element that's not <rdf:Seq>
889 private function startElementModeSeq( $elm ) {
890 if ( $elm === self
::NS_RDF
. ' Seq' ) {
891 array_unshift( $this->mode
, self
::MODE_LI
);
892 } elseif ( $elm === self
::NS_RDF
. ' Bag' ) {
894 $this->logger
->info( __METHOD__
. ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
895 . ' it is a Seq, since some buggy software is known to screw this up.' );
896 array_unshift( $this->mode
, self
::MODE_LI
);
898 throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
903 * Start element in MODE_LANG (language alternative)
904 * this should always be <rdf:Alt>
906 * This tag tends to be used for metadata like describe this
907 * picture, which can be translated into multiple languages.
909 * XMP supports non-linguistic alternative selections,
910 * which are really only used for thumbnails, which
911 * we don't care about.
913 * @param string $elm Namespace . ' ' . tag
914 * @throws RuntimeException If we have an element that's not <rdf:Alt>
916 private function startElementModeLang( $elm ) {
917 if ( $elm === self
::NS_RDF
. ' Alt' ) {
918 array_unshift( $this->mode
, self
::MODE_LI_LANG
);
920 throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
925 * Handle an opening element when in MODE_SIMPLE
927 * This should not happen often. This is for if a simple element
928 * already opened has a child element. Could happen for a
932 * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
933 * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
934 * </exif:DigitalZoomRatio>
936 * This method is called when processing the <rdf:Description> element
938 * @param string $elm Namespace and tag names separated by space.
939 * @param array $attribs Attributes of the element.
940 * @throws RuntimeException
942 private function startElementModeSimple( $elm, $attribs ) {
943 if ( $elm === self
::NS_RDF
. ' Description' ) {
944 // If this value has qualifiers
945 array_unshift( $this->mode
, self
::MODE_QDESC
);
946 array_unshift( $this->curItem
, $this->curItem
[0] );
948 if ( isset( $attribs[self
::NS_RDF
. ' value'] ) ) {
949 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
950 $this->saveValue( $ns, $tag, $attribs[self
::NS_RDF
. ' value'] );
952 } elseif ( $elm === self
::NS_RDF
. ' value' ) {
953 // This should not be here.
954 throw new RuntimeException( __METHOD__
. ' Encountered <rdf:value> where it was unexpected.' );
956 // something else we don't recognize, like a qualifier maybe.
957 $this->logger
->info( __METHOD__
.
958 " Encountered element <{element}> where only expecting character data as value of {curitem}",
959 [ 'element' => $elm, 'curitem' => $this->curItem
[0] ] );
960 array_unshift( $this->mode
, self
::MODE_IGNORE
);
961 array_unshift( $this->curItem
, $elm );
966 * Start an element when in MODE_QDESC.
967 * This generally happens when a simple element has an inner
968 * rdf:Description to hold qualifier elements.
971 * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
972 * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
973 * </exif:DigitalZoomRatio>
974 * Called when processing the <rdf:value> or <foo:someQualifier>.
976 * @param string $elm Namespace and tag name separated by a space.
978 private function startElementModeQDesc( $elm ) {
979 if ( $elm === self
::NS_RDF
. ' value' ) {
980 return; // do nothing
982 // otherwise its a qualifier, which we ignore
983 array_unshift( $this->mode
, self
::MODE_IGNORE
);
984 array_unshift( $this->curItem
, $elm );
989 * Starting an element when in MODE_INITIAL
990 * This usually happens when we hit an element inside
991 * the outer rdf:Description
993 * This is generally where most properties start.
995 * @param string $ns Namespace
996 * @param string $tag Tag name (without namespace prefix)
997 * @param array $attribs Array of attributes
998 * @throws RuntimeException
1000 private function startElementModeInitial( $ns, $tag, $attribs ) {
1001 if ( $ns !== self
::NS_RDF
) {
1002 if ( isset( $this->items
[$ns][$tag] ) ) {
1003 if ( isset( $this->items
[$ns][$tag]['structPart'] ) ) {
1004 // If this element is supposed to appear only as
1005 // a child of a structure, but appears here (not as
1006 // a child of a struct), then something weird is
1007 // happening, so ignore this element and its children.
1009 $this->logger
->warning(
1010 'Encountered <{element}> outside of its expected parent. Ignoring.',
1011 [ 'element' => "$ns:$tag" ]
1014 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1015 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1019 $mode = $this->items
[$ns][$tag]['mode'];
1020 array_unshift( $this->mode
, $mode );
1021 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1022 if ( $mode === self
::MODE_STRUCT
) {
1023 $this->ancestorStruct
= isset( $this->items
[$ns][$tag]['map_name'] )
1024 ?
$this->items
[$ns][$tag]['map_name'] : $tag;
1026 if ( $this->charContent
!== false ) {
1028 // Should not happen in valid XMP.
1029 throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1032 // This element is not on our list of allowed elements so ignore.
1033 $this->logger
->debug( __METHOD__
. ' Ignoring unrecognized element <{element}>.',
1034 [ 'element' => "$ns:$tag" ] );
1035 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1036 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1041 // process attributes
1042 $this->doAttribs( $attribs );
1046 * Hit an opening element when in a Struct (MODE_STRUCT)
1047 * This is generally for fields of a compound property.
1049 * Example of a struct (abbreviated; flash has more properties):
1051 * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1052 * <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1056 * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1057 * <exif:Mode>1</exif:Mode></exif:Flash>
1059 * @param string $ns Namespace
1060 * @param string $tag Tag name (no ns)
1061 * @param array $attribs Array of attribs w/ values.
1062 * @throws RuntimeException
1064 private function startElementModeStruct( $ns, $tag, $attribs ) {
1065 if ( $ns !== self
::NS_RDF
) {
1066 if ( isset( $this->items
[$ns][$tag] ) ) {
1067 if ( isset( $this->items
[$ns][$this->ancestorStruct
]['children'] )
1068 && !isset( $this->items
[$ns][$this->ancestorStruct
]['children'][$tag] )
1070 // This assumes that we don't have inter-namespace nesting
1071 // which we don't in all the properties we're interested in.
1072 throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1073 . "> where it is not allowed." );
1075 array_unshift( $this->mode
, $this->items
[$ns][$tag]['mode'] );
1076 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1077 if ( $this->charContent
!== false ) {
1079 // Should not happen in valid XMP.
1080 throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1081 $this->charContent
. ")." );
1084 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1085 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1091 if ( $ns === self
::NS_RDF
&& $tag === 'Description' ) {
1092 $this->doAttribs( $attribs );
1093 array_unshift( $this->mode
, self
::MODE_STRUCT
);
1094 array_unshift( $this->curItem
, $this->curItem
[0] );
1099 * opening element in MODE_LI
1100 * process elements of arrays.
1103 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1104 * </rdf:Seq> </exif:ISOSpeedRatings>
1105 * This method is called when we hit the <rdf:li> element.
1107 * @param string $elm Namespace . ' ' . tagname
1108 * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1109 * @throws RuntimeException If gets a tag other than <rdf:li>
1111 private function startElementModeLi( $elm, $attribs ) {
1112 if ( ( $elm ) !== self
::NS_RDF
. ' li' ) {
1113 throw new RuntimeException( "<rdf:li> expected but got $elm." );
1116 if ( !isset( $this->mode
[1] ) ) {
1117 // This should never ever ever happen. Checking for it
1119 throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1122 if ( $this->mode
[1] === self
::MODE_BAGSTRUCT
) {
1123 // This list item contains a compound (STRUCT) value.
1124 array_unshift( $this->mode
, self
::MODE_STRUCT
);
1125 array_unshift( $this->curItem
, $elm );
1126 $this->processingArray
= true;
1128 if ( !isset( $this->curItem
[1] ) ) {
1130 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1132 list( $curNS, $curTag ) = explode( ' ', $this->curItem
[1] );
1133 $this->ancestorStruct
= isset( $this->items
[$curNS][$curTag]['map_name'] )
1134 ?
$this->items
[$curNS][$curTag]['map_name'] : $curTag;
1136 $this->doAttribs( $attribs );
1138 // Normal BAG or SEQ containing simple values.
1139 array_unshift( $this->mode
, self
::MODE_SIMPLE
);
1140 // need to add curItem[0] on again since one is for the specific item
1141 // and one is for the entire group.
1142 array_unshift( $this->curItem
, $this->curItem
[0] );
1143 $this->processingArray
= true;
1148 * Opening element in MODE_LI_LANG.
1149 * process elements of language alternatives
1152 * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1153 * </rdf:li> </rdf:Alt> </dc:title>
1155 * This method is called when we hit the <rdf:li> element.
1157 * @param string $elm Namespace . ' ' . tag
1158 * @param array $attribs Array of elements (most importantly xml:lang)
1159 * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1161 private function startElementModeLiLang( $elm, $attribs ) {
1162 if ( $elm !== self
::NS_RDF
. ' li' ) {
1163 throw new RuntimeException( __METHOD__
. " <rdf:li> expected but got $elm." );
1165 if ( !isset( $attribs[self
::NS_XML
. ' lang'] )
1166 ||
!preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self
::NS_XML
. ' lang'] )
1168 throw new RuntimeException( __METHOD__
1169 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1172 // Lang is case-insensitive.
1173 $this->itemLang
= strtolower( $attribs[self
::NS_XML
. ' lang'] );
1175 // need to add curItem[0] on again since one is for the specific item
1176 // and one is for the entire group.
1177 array_unshift( $this->curItem
, $this->curItem
[0] );
1178 array_unshift( $this->mode
, self
::MODE_SIMPLE
);
1179 $this->processingArray
= true;
1183 * Hits an opening element.
1184 * Generally just calls a helper based on what MODE we're in.
1185 * Also does some initial set up for the wrapper element
1187 * @param resource $parser
1188 * @param string $elm Namespace "<space>" element
1189 * @param array $attribs Attribute name => value
1190 * @throws RuntimeException
1192 function startElement( $parser, $elm, $attribs ) {
1193 if ( $elm === self
::NS_RDF
. ' RDF'
1194 ||
$elm === 'adobe:ns:meta/ xmpmeta'
1195 ||
$elm === 'adobe:ns:meta/ xapmeta'
1199 } elseif ( $elm === self
::NS_RDF
. ' Description' ) {
1200 if ( count( $this->mode
) === 0 ) {
1202 array_unshift( $this->mode
, self
::MODE_INITIAL
);
1204 } elseif ( $elm === self
::NS_RDF
. ' type' ) {
1205 // This doesn't support rdf:type properly.
1206 // In practise I have yet to see a file that
1207 // uses this element, however it is mentioned
1208 // on page 25 of part 1 of the xmp standard.
1209 // Also it seems as if exiv2 and exiftool do not support
1210 // this either (That or I misunderstand the standard)
1211 $this->logger
->info( __METHOD__
. ' Encountered <rdf:type> which isn\'t currently supported' );
1214 if ( strpos( $elm, ' ' ) === false ) {
1215 // This probably shouldn't happen.
1216 $this->logger
->info( __METHOD__
. " Encountered <$elm> which has no namespace. Skipping." );
1221 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1223 if ( count( $this->mode
) === 0 ) {
1224 // This should not happen.
1225 throw new RuntimeException( 'Error extracting XMP, '
1226 . "encountered <$elm> with no mode" );
1229 switch ( $this->mode
[0] ) {
1230 case self
::MODE_IGNORE
:
1231 $this->startElementModeIgnore( $elm );
1233 case self
::MODE_SIMPLE
:
1234 $this->startElementModeSimple( $elm, $attribs );
1236 case self
::MODE_INITIAL
:
1237 $this->startElementModeInitial( $ns, $tag, $attribs );
1239 case self
::MODE_STRUCT
:
1240 $this->startElementModeStruct( $ns, $tag, $attribs );
1242 case self
::MODE_BAG
:
1243 case self
::MODE_BAGSTRUCT
:
1244 $this->startElementModeBag( $elm );
1246 case self
::MODE_SEQ
:
1247 $this->startElementModeSeq( $elm );
1249 case self
::MODE_LANG
:
1250 $this->startElementModeLang( $elm );
1252 case self
::MODE_LI_LANG
:
1253 $this->startElementModeLiLang( $elm, $attribs );
1256 $this->startElementModeLi( $elm, $attribs );
1258 case self
::MODE_QDESC
:
1259 $this->startElementModeQDesc( $elm );
1262 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode
[0] );
1266 // @codingStandardsIgnoreStart Generic.Files.LineLength
1268 * Process attributes.
1269 * Simple values can be stored as either a tag or attribute
1271 * Often the initial "<rdf:Description>" tag just has all the simple
1272 * properties as attributes.
1276 * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1279 * @param array $attribs Array attribute=>value
1280 * @throws RuntimeException
1282 // @codingStandardsIgnoreEnd
1283 private function doAttribs( $attribs ) {
1284 // first check for rdf:parseType attribute, as that can change
1285 // how the attributes are interperted.
1287 if ( isset( $attribs[self
::NS_RDF
. ' parseType'] )
1288 && $attribs[self
::NS_RDF
. ' parseType'] === 'Resource'
1289 && $this->mode
[0] === self
::MODE_SIMPLE
1291 // this is equivalent to having an inner rdf:Description
1292 $this->mode
[0] = self
::MODE_QDESC
;
1294 foreach ( $attribs as $name => $val ) {
1295 if ( strpos( $name, ' ' ) === false ) {
1296 // This shouldn't happen, but so far some old software forgets namespace
1298 $this->logger
->info( __METHOD__
. ' Encountered non-namespaced attribute: '
1299 . " $name=\"$val\". Skipping. " );
1302 list( $ns, $tag ) = explode( ' ', $name, 2 );
1303 if ( $ns === self
::NS_RDF
) {
1304 if ( $tag === 'value' ||
$tag === 'resource' ) {
1305 // resource is for url.
1306 // value attribute is a weird way of just putting the contents.
1307 $this->char( $this->xmlParser
, $val );
1309 } elseif ( isset( $this->items
[$ns][$tag] ) ) {
1310 if ( $this->mode
[0] === self
::MODE_SIMPLE
) {
1311 throw new RuntimeException( __METHOD__
1312 . " $ns:$tag found as attribute where not allowed" );
1314 $this->saveValue( $ns, $tag, $val );
1316 $this->logger
->debug( __METHOD__
. " Ignoring unrecognized element <$ns:$tag>." );
1322 * Given an extracted value, save it to results array
1324 * note also uses $this->ancestorStruct and
1325 * $this->processingArray to determine what name to
1326 * save the value under. (in addition to $tag).
1328 * @param string $ns Namespace of tag this is for
1329 * @param string $tag Tag name
1330 * @param string $val Value to save
1332 private function saveValue( $ns, $tag, $val ) {
1333 $info =& $this->items
[$ns][$tag];
1334 $finalName = isset( $info['map_name'] )
1335 ?
$info['map_name'] : $tag;
1336 if ( isset( $info['validate'] ) ) {
1337 if ( is_array( $info['validate'] ) ) {
1338 $validate = $info['validate'];
1340 $validator = new XMPValidate( $this->logger
);
1341 $validate = [ $validator, $info['validate'] ];
1344 if ( is_callable( $validate ) ) {
1345 call_user_func_array( $validate, [ $info, &$val, true ] );
1346 // the reasoning behind using &$val instead of using the return value
1347 // is to be consistent between here and validating structures.
1348 if ( is_null( $val ) ) {
1349 $this->logger
->info( __METHOD__
. " <$ns:$tag> failed validation." );
1354 $this->logger
->warning( __METHOD__
. " Validation function for $finalName ("
1355 . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1359 if ( $this->ancestorStruct
&& $this->processingArray
) {
1360 // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1361 $this->results
['xmp-' . $info['map_group']][$this->ancestorStruct
][][$finalName] = $val;
1362 } elseif ( $this->ancestorStruct
) {
1363 $this->results
['xmp-' . $info['map_group']][$this->ancestorStruct
][$finalName] = $val;
1364 } elseif ( $this->processingArray
) {
1365 if ( $this->itemLang
=== false ) {
1367 $this->results
['xmp-' . $info['map_group']][$finalName][] = $val;
1370 $this->results
['xmp-' . $info['map_group']][$finalName][$this->itemLang
] = $val;
1373 $this->results
['xmp-' . $info['map_group']][$finalName] = $val;