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
;
29 * Class for reading xmp data containing properties relevant to
30 * images, and spitting out an array that FormatMetadata accepts.
32 * Note, this is not meant to recognize every possible thing you can
33 * encode in XMP. It should recognize all the properties we want.
34 * For example it doesn't have support for structures with multiple
35 * nesting levels, as none of the properties we're supporting use that
36 * feature. If it comes across properties it doesn't recognize, it should
39 * The public methods one would call in this class are
41 * Reads in xmp content.
42 * Can potentially be called multiple times with partial data each time.
43 * - parseExtended( $content )
44 * Reads XMPExtended blocks (jpeg files only).
46 * Outputs a results array.
48 * Note XMP kind of looks like rdf. They are not the same thing - XMP is
49 * encoded as a specific subset of rdf. This class can read XMP. It cannot
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 = array();
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) */
67 private $mode = array();
69 /** @var array Array to hold results */
70 private $results = array();
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
135 * Primary job is to initialize the XMLParser
137 function __construct( LoggerInterface
$logger = null ) {
139 if ( !function_exists( 'xml_parser_create_ns' ) ) {
140 // this should already be checked by this point
141 throw new RuntimeException( 'XMP support requires XML Parser' );
144 $this->setLogger( $logger );
146 $this->setLogger( new NullLogger() );
149 $this->items
= XMPInfo
::getItems();
151 $this->resetXMLParser();
154 public function setLogger( LoggerInterface
$logger ) {
155 $this->logger
= $logger;
159 * free the XML parser.
161 * @note It is unclear to me if we really need to do this ourselves
162 * or if php garbage collection will automatically free the xmlParser
163 * when it is no longer needed.
165 private function destroyXMLParser() {
166 if ( $this->xmlParser
) {
167 xml_parser_free( $this->xmlParser
);
168 $this->xmlParser
= null;
173 * Main use is if a single item has multiple xmp documents describing it.
174 * For example in jpeg's with extendedXMP
176 private function resetXMLParser() {
178 $this->destroyXMLParser();
180 $this->xmlParser
= xml_parser_create_ns( 'UTF-8', ' ' );
181 xml_parser_set_option( $this->xmlParser
, XML_OPTION_CASE_FOLDING
, 0 );
182 xml_parser_set_option( $this->xmlParser
, XML_OPTION_SKIP_WHITE
, 1 );
184 xml_set_element_handler( $this->xmlParser
,
185 array( $this, 'startElement' ),
186 array( $this, 'endElement' ) );
188 xml_set_character_data_handler( $this->xmlParser
, array( $this, 'char' ) );
190 $this->parsable
= self
::PARSABLE_UNKNOWN
;
191 $this->xmlParsableBuffer
= '';
195 * Check if this instance supports using this class
197 public static function isSupported() {
198 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
201 /** Get the result array. Do some post-processing before returning
202 * the array, and transform any metadata that is special-cased.
204 * @return array Array of results as an array of arrays suitable for
205 * FormatMetadata::getFormattedData().
207 public function getResults() {
208 // xmp-special is for metadata that affects how stuff
209 // is extracted. For example xmpNote:HasExtendedXMP.
211 // It is also used to handle photoshop:AuthorsPosition
212 // which is weird and really part of another property,
213 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
214 // The location fields also use it.
216 $data = $this->results
;
218 if ( isset( $data['xmp-special']['AuthorsPosition'] )
219 && is_string( $data['xmp-special']['AuthorsPosition'] )
220 && isset( $data['xmp-general']['Artist'][0] )
222 // Note, if there is more than one creator,
223 // this only applies to first. This also will
224 // only apply to the dc:Creator prop, not the
227 $data['xmp-general']['Artist'][0] =
228 $data['xmp-special']['AuthorsPosition'] . ', '
229 . $data['xmp-general']['Artist'][0];
232 // Go through the LocationShown and LocationCreated
233 // changing it to the non-hierarchal form used by
234 // the other location fields.
236 if ( isset( $data['xmp-special']['LocationShown'][0] )
237 && is_array( $data['xmp-special']['LocationShown'][0] )
239 // the is_array is just paranoia. It should always
241 foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
242 if ( !is_array( $loc ) ) {
243 // To avoid copying over the _type meta-fields.
246 foreach ( $loc as $field => $val ) {
247 $data['xmp-general'][$field . 'Dest'][] = $val;
251 if ( isset( $data['xmp-special']['LocationCreated'][0] )
252 && is_array( $data['xmp-special']['LocationCreated'][0] )
254 // the is_array is just paranoia. It should always
256 foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
257 if ( !is_array( $loc ) ) {
258 // To avoid copying over the _type meta-fields.
261 foreach ( $loc as $field => $val ) {
262 $data['xmp-general'][$field . 'Created'][] = $val;
267 // We don't want to return the special values, since they're
268 // special and not info to be stored about the file.
269 unset( $data['xmp-special'] );
271 // Convert GPSAltitude to negative if below sea level.
272 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
273 && isset( $data['xmp-exif']['GPSAltitude'] )
276 // Must convert to a real before multiplying by -1
277 // XMPValidate guarantees there will always be a '/' in this value.
278 list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
279 $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
281 if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
282 $data['xmp-exif']['GPSAltitude'] *= -1;
284 unset( $data['xmp-exif']['GPSAltitudeRef'] );
291 * Main function to call to parse XMP. Use getResults to
294 * Also catches any errors during processing, writes them to
295 * debug log, blanks result array and returns false.
297 * @param string $content XMP data
298 * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
299 * @throws RuntimeException
300 * @return bool Success.
302 public function parse( $content, $allOfIt = true ) {
303 if ( !$this->xmlParser
) {
304 $this->resetXMLParser();
308 // detect encoding by looking for BOM which is supposed to be in processing instruction.
309 // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
310 if ( !$this->charset
) {
312 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
317 $this->charset
= 'UTF-16BE';
320 $this->charset
= 'UTF-16LE';
322 case "\x00\x00\xFE\xFF":
323 $this->charset
= 'UTF-32BE';
325 case "\xFF\xFE\x00\x00":
326 $this->charset
= 'UTF-32LE';
329 $this->charset
= 'UTF-8';
332 //this should be impossible to get to
333 throw new RuntimeException( "Invalid BOM" );
336 // standard specifically says, if no bom assume utf-8
337 $this->charset
= 'UTF-8';
340 if ( $this->charset
!== 'UTF-8' ) {
341 //don't convert if already utf-8
342 MediaWiki\
suppressWarnings();
343 $content = iconv( $this->charset
, 'UTF-8//IGNORE', $content );
344 MediaWiki\restoreWarnings
();
347 // Ensure the XMP block does not have an xml doctype declaration, which
348 // could declare entities unsafe to parse with xml_parse (T85848/T71210).
349 if ( $this->parsable
!== self
::PARSABLE_OK
) {
350 if ( $this->parsable
=== self
::PARSABLE_NO
) {
351 throw new RuntimeException( 'Unsafe doctype declaration in XML.' );
354 $content = $this->xmlParsableBuffer
. $content;
355 if ( !$this->checkParseSafety( $content ) ) {
356 if ( !$allOfIt && $this->parsable
!== self
::PARSABLE_NO
) {
357 // parse wasn't Unsuccessful yet, so return true
361 $msg = ( $this->parsable
=== self
::PARSABLE_NO
) ?
362 'Unsafe doctype declaration in XML.' :
363 'No root element found in XML.';
364 throw new RuntimeException( $msg );
368 $ok = xml_parse( $this->xmlParser
, $content, $allOfIt );
370 $code = xml_get_error_code( $this->xmlParser
);
371 $error = xml_error_string( $code );
372 $line = xml_get_current_line_number( $this->xmlParser
);
373 $col = xml_get_current_column_number( $this->xmlParser
);
374 $offset = xml_get_current_byte_index( $this->xmlParser
);
376 $this->logger
->warning(
377 '{method} : Error reading XMP content: {error} ' .
378 '(line: {line} column: {column} byte offset: {offset})',
380 'method' => __METHOD__
,
381 'error_code' => $code,
386 'content' => $content,
388 $this->results
= array(); // blank if error.
389 $this->destroyXMLParser();
392 } catch ( Exception
$e ) {
393 $this->logger
->warning(
394 '{method} Exception caught while parsing: ' . $e->getMessage(),
396 'method' => __METHOD__
,
398 'content' => $content,
401 $this->results
= array();
405 $this->destroyXMLParser();
411 /** Entry point for XMPExtended blocks in jpeg files
413 * @todo In serious need of testing
414 * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
415 * @param string $content XMPExtended block minus the namespace signature
416 * @return bool If it succeeded.
418 public function parseExtended( $content ) {
419 // @todo FIXME: This is untested. Hard to find example files
420 // or programs that make such files..
421 $guid = substr( $content, 0, 32 );
422 if ( !isset( $this->results
['xmp-special']['HasExtendedXMP'] )
423 ||
$this->results
['xmp-special']['HasExtendedXMP'] !== $guid
425 $this->logger
->info( __METHOD__
.
426 " Ignoring XMPExtended block due to wrong guid (guid= '$guid')" );
430 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
432 if ( !$len ||
$len['length'] < 4 ||
$len['offset'] < 0 ||
$len['offset'] > $len['length'] ) {
433 $this->logger
->info( __METHOD__
. 'Error reading extended XMP block, invalid length or offset.' );
438 // we're not very robust here. we should accept it in the wrong order.
439 // To quote the XMP standard:
440 // "A JPEG writer should write the ExtendedXMP marker segments in order,
441 // immediately following the StandardXMP. However, the JPEG standard
442 // does not require preservation of marker segment order. A robust JPEG
443 // reader should tolerate the marker segments in any order."
445 // otoh the probability that an image will have more than 128k of
446 // metadata is rather low... so the probability that it will have
447 // > 128k, and be in the wrong order is very low...
449 if ( $len['offset'] !== $this->extendedXMPOffset
) {
450 $this->logger
->info( __METHOD__
. 'Ignoring XMPExtended block due to wrong order. (Offset was '
451 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset
. ')' );
456 if ( $len['offset'] === 0 ) {
457 // if we're starting the extended block, we've probably already
458 // done the XMPStandard block, so reset.
459 $this->resetXMLParser();
462 $this->extendedXMPOffset +
= $len['length'];
464 $actualContent = substr( $content, 40 );
466 if ( $this->extendedXMPOffset
=== strlen( $actualContent ) ) {
472 $this->logger
->debug( __METHOD__
. 'Parsing a XMPExtended block' );
474 return $this->parse( $actualContent, $atEnd );
478 * Character data handler
479 * Called whenever character data is found in the xmp document.
481 * does nothing if we're in MODE_IGNORE or if the data is whitespace
482 * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
483 * data in the other modes).
485 * As an example, this happens when we encounter XMP like:
486 * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
487 * and are processing the 0/10 bit.
489 * @param XMLParser $parser XMLParser reference to the xml parser
490 * @param string $data Character data
491 * @throws RuntimeException On invalid data
493 function char( $parser, $data ) {
495 $data = trim( $data );
496 if ( trim( $data ) === "" ) {
500 if ( !isset( $this->mode
[0] ) ) {
501 throw new RuntimeException( 'Unexpected character data before first rdf:Description element' );
504 if ( $this->mode
[0] === self
::MODE_IGNORE
) {
508 if ( $this->mode
[0] !== self
::MODE_SIMPLE
509 && $this->mode
[0] !== self
::MODE_QDESC
511 throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode
[0] . ')' );
514 // to check, how does this handle w.s.
515 if ( $this->charContent
=== false ) {
516 $this->charContent
= $data;
518 $this->charContent
.= $data;
523 * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
524 * contain a doctype declaration which could contain a dos attack if we
525 * parse it and expand internal entities (T85848).
527 * @param string $content xml string to check for parse safety
528 * @return bool true if the xml is safe to parse, false otherwise
530 private function checkParseSafety( $content ) {
531 $reader = new XMLReader();
534 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
535 // instead of using XML().
537 'data://text/plain,' . urlencode( $content ),
539 LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
542 $oldDisable = libxml_disable_entity_loader( true );
543 /** @noinspection PhpUnusedLocalVariableInspection */
544 $reset = new ScopedCallback(
545 'libxml_disable_entity_loader',
548 $reader->setParserProperty( XMLReader
::SUBST_ENTITIES
, false );
550 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
551 // when parsing truncated XML, which causes unit tests to fail.
552 MediaWiki\
suppressWarnings();
553 while ( $reader->read() ) {
554 if ( $reader->nodeType
=== XMLReader
::ELEMENT
) {
555 // Reached the first element without hitting a doctype declaration
556 $this->parsable
= self
::PARSABLE_OK
;
560 if ( $reader->nodeType
=== XMLReader
::DOC_TYPE
) {
561 $this->parsable
= self
::PARSABLE_NO
;
566 MediaWiki\restoreWarnings
();
568 if ( !is_null( $result ) ) {
572 // Reached the end of the parsable xml without finding an element
573 // or doctype. Buffer and try again.
574 $this->parsable
= self
::PARSABLE_BUFFERING
;
575 $this->xmlParsableBuffer
= $content;
579 /** When we hit a closing element in MODE_IGNORE
580 * Check to see if this is the element we started to ignore,
581 * in which case we get out of MODE_IGNORE
583 * @param string $elm Namespace of element followed by a space and then tag name of element.
585 private function endElementModeIgnore( $elm ) {
586 if ( $this->curItem
[0] === $elm ) {
587 array_shift( $this->curItem
);
588 array_shift( $this->mode
);
593 * Hit a closing element when in MODE_SIMPLE.
594 * This generally means that we finished processing a
595 * property value, and now have to save the result to the
598 * For example, when processing:
599 * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
600 * this deals with when we hit </exif:DigitalZoomRatio>.
602 * Or it could be if we hit the end element of a property
603 * of a compound data structure (like a member of an array).
605 * @param string $elm Namespace, space, and tag name.
607 private function endElementModeSimple( $elm ) {
608 if ( $this->charContent
!== false ) {
609 if ( $this->processingArray
) {
610 // if we're processing an array, use the original element
611 // name instead of rdf:li.
612 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
614 list( $ns, $tag ) = explode( ' ', $elm, 2 );
616 $this->saveValue( $ns, $tag, $this->charContent
);
618 $this->charContent
= false; // reset
620 array_shift( $this->curItem
);
621 array_shift( $this->mode
);
625 * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
626 * generally means we've finished processing a nested structure.
627 * resets some internal variables to indicate that.
629 * Note this means we hit the closing element not the "</rdf:Seq>".
631 * @par For example, when processing:
633 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
634 * </rdf:Seq> </exif:ISOSpeedRatings>
637 * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
639 * @param string $elm Namespace . space . tag name.
640 * @throws RuntimeException
642 private function endElementNested( $elm ) {
644 /* cur item must be the same as $elm, unless if in MODE_STRUCT
645 in which case it could also be rdf:Description */
646 if ( $this->curItem
[0] !== $elm
647 && !( $elm === self
::NS_RDF
. ' Description'
648 && $this->mode
[0] === self
::MODE_STRUCT
)
650 throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" .
651 $this->curItem
[0] . '>' );
654 // Validate structures.
655 list( $ns, $tag ) = explode( ' ', $elm, 2 );
656 if ( isset( $this->items
[$ns][$tag]['validate'] ) ) {
657 $info =& $this->items
[$ns][$tag];
658 $finalName = isset( $info['map_name'] )
659 ?
$info['map_name'] : $tag;
661 if ( is_array( $info['validate'] ) ) {
662 $validate = $info['validate'];
664 $validator = new XMPValidate( $this->logger
);
665 $validate = array( $validator, $info['validate'] );
668 if ( !isset( $this->results
['xmp-' . $info['map_group']][$finalName] ) ) {
669 // This can happen if all the members of the struct failed validation.
670 $this->logger
->debug( __METHOD__
. " <$ns:$tag> has no valid members." );
671 } elseif ( is_callable( $validate ) ) {
672 $val =& $this->results
['xmp-' . $info['map_group']][$finalName];
673 call_user_func_array( $validate, array( $info, &$val, false ) );
674 if ( is_null( $val ) ) {
675 // the idea being the validation function will unset the variable if
677 $this->logger
->info( __METHOD__
. " <$ns:$tag> failed validation." );
678 unset( $this->results
['xmp-' . $info['map_group']][$finalName] );
681 $this->logger
->warning( __METHOD__
. " Validation function for $finalName ("
682 . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
686 array_shift( $this->curItem
);
687 array_shift( $this->mode
);
688 $this->ancestorStruct
= false;
689 $this->processingArray
= false;
690 $this->itemLang
= false;
694 * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
695 * Add information about what type of element this is.
697 * Note we still have to hit the outer "</property>"
699 * @par For example, when processing:
701 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
702 * </rdf:Seq> </exif:ISOSpeedRatings>
705 * This method is called when we hit the "</rdf:Seq>".
706 * (For comparison, we call endElementModeSimple when we
707 * hit the "</rdf:li>")
709 * @param string $elm Namespace . ' ' . element name
710 * @throws RuntimeException
712 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( __METHOD__
. " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
742 * End element while in MODE_QDESC
743 * mostly when ending an element when we have a simple value
744 * that has qualifiers.
746 * Qualifiers aren't all that common, and we don't do anything
749 * @param string $elm Namespace and element
751 private function endElementModeQDesc( $elm ) {
753 if ( $elm === self
::NS_RDF
. ' value' ) {
754 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
755 $this->saveValue( $ns, $tag, $this->charContent
);
759 array_shift( $this->mode
);
760 array_shift( $this->curItem
);
765 * Handler for hitting a closing element.
767 * generally just calls a helper function depending on what
770 * Ignores the outer wrapping elements that are optional in
771 * xmp and have no meaning.
773 * @param XMLParser $parser
774 * @param string $elm Namespace . ' ' . element name
775 * @throws RuntimeException
777 function endElement( $parser, $elm ) {
778 if ( $elm === ( self
::NS_RDF
. ' RDF' )
779 ||
$elm === 'adobe:ns:meta/ xmpmeta'
780 ||
$elm === 'adobe:ns:meta/ xapmeta'
786 if ( $elm === self
::NS_RDF
. ' type' ) {
787 // these aren't really supported properly yet.
788 // However, it appears they almost never used.
789 $this->logger
->info( __METHOD__
. ' encountered <rdf:type>' );
792 if ( strpos( $elm, ' ' ) === false ) {
793 // This probably shouldn't happen.
794 // However, there is a bug in an adobe product
795 // that forgets the namespace on some things.
796 // (Luckily they are unimportant things).
797 $this->logger
->info( __METHOD__
. " Encountered </$elm> which has no namespace. Skipping." );
802 if ( count( $this->mode
[0] ) === 0 ) {
803 // This should never ever happen and means
804 // there is a pretty major bug in this class.
805 throw new RuntimeException( 'Encountered end element with no mode' );
808 if ( count( $this->curItem
) == 0 && $this->mode
[0] !== self
::MODE_INITIAL
) {
809 // just to be paranoid. Should always have a curItem, except for initially
810 // (aka during MODE_INITAL).
811 throw new RuntimeException( "Hit end element </$elm> but no curItem" );
814 switch ( $this->mode
[0] ) {
815 case self
::MODE_IGNORE
:
816 $this->endElementModeIgnore( $elm );
818 case self
::MODE_SIMPLE
:
819 $this->endElementModeSimple( $elm );
821 case self
::MODE_STRUCT
:
824 case self
::MODE_LANG
:
825 case self
::MODE_BAGSTRUCT
:
826 $this->endElementNested( $elm );
828 case self
::MODE_INITIAL
:
829 if ( $elm === self
::NS_RDF
. ' Description' ) {
830 array_shift( $this->mode
);
832 throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' );
836 case self
::MODE_LI_LANG
:
837 $this->endElementModeLi( $elm );
839 case self
::MODE_QDESC
:
840 $this->endElementModeQDesc( $elm );
843 $this->logger
->warning( __METHOD__
. " no mode (elm = $elm)" );
849 * Hit an opening element while in MODE_IGNORE
851 * XMP is extensible, so ignore any tag we don't understand.
853 * Mostly ignores, unless we encounter the element that we are ignoring.
854 * in which case we add it to the item stack, so we can ignore things
855 * that are nested, correctly.
857 * @param string $elm Namespace . ' ' . tag name
859 private function startElementModeIgnore( $elm ) {
860 if ( $elm === $this->curItem
[0] ) {
861 array_unshift( $this->curItem
, $elm );
862 array_unshift( $this->mode
, self
::MODE_IGNORE
);
867 * Start element in MODE_BAG (unordered array)
868 * this should always be <rdf:Bag>
870 * @param string $elm Namespace . ' ' . tag
871 * @throws RuntimeException If we have an element that's not <rdf:Bag>
873 private function startElementModeBag( $elm ) {
874 if ( $elm === self
::NS_RDF
. ' Bag' ) {
875 array_unshift( $this->mode
, self
::MODE_LI
);
877 throw new RuntimeException( "Expected <rdf:Bag> but got $elm." );
882 * Start element in MODE_SEQ (ordered array)
883 * this should always be <rdf:Seq>
885 * @param string $elm Namespace . ' ' . tag
886 * @throws RuntimeException If we have an element that's not <rdf:Seq>
888 private function startElementModeSeq( $elm ) {
889 if ( $elm === self
::NS_RDF
. ' Seq' ) {
890 array_unshift( $this->mode
, self
::MODE_LI
);
891 } elseif ( $elm === self
::NS_RDF
. ' Bag' ) {
893 $this->logger
->info( __METHOD__
. ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
894 . ' it is a Seq, since some buggy software is known to screw this up.' );
895 array_unshift( $this->mode
, self
::MODE_LI
);
897 throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
902 * Start element in MODE_LANG (language alternative)
903 * this should always be <rdf:Alt>
905 * This tag tends to be used for metadata like describe this
906 * picture, which can be translated into multiple languages.
908 * XMP supports non-linguistic alternative selections,
909 * which are really only used for thumbnails, which
910 * we don't care about.
912 * @param string $elm Namespace . ' ' . tag
913 * @throws RuntimeException If we have an element that's not <rdf:Alt>
915 private function startElementModeLang( $elm ) {
916 if ( $elm === self
::NS_RDF
. ' Alt' ) {
917 array_unshift( $this->mode
, self
::MODE_LI_LANG
);
919 throw new RuntimeException( "Expected <rdf:Seq> but got $elm." );
924 * Handle an opening element when in MODE_SIMPLE
926 * This should not happen often. This is for if a simple element
927 * already opened has a child element. Could happen for a
931 * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
932 * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
933 * </exif:DigitalZoomRatio>
935 * This method is called when processing the <rdf:Description> element
937 * @param string $elm Namespace and tag names separated by space.
938 * @param array $attribs Attributes of the element.
939 * @throws RuntimeException
941 private function startElementModeSimple( $elm, $attribs ) {
942 if ( $elm === self
::NS_RDF
. ' Description' ) {
943 // If this value has qualifiers
944 array_unshift( $this->mode
, self
::MODE_QDESC
);
945 array_unshift( $this->curItem
, $this->curItem
[0] );
947 if ( isset( $attribs[self
::NS_RDF
. ' value'] ) ) {
948 list( $ns, $tag ) = explode( ' ', $this->curItem
[0], 2 );
949 $this->saveValue( $ns, $tag, $attribs[self
::NS_RDF
. ' value'] );
951 } elseif ( $elm === self
::NS_RDF
. ' value' ) {
952 // This should not be here.
953 throw new RuntimeException( __METHOD__
. ' Encountered <rdf:value> where it was unexpected.' );
955 // something else we don't recognize, like a qualifier maybe.
956 $this->logger
->info( __METHOD__
.
957 " Encountered element <$elm> where only expecting character data as value of " .
959 array_unshift( $this->mode
, self
::MODE_IGNORE
);
960 array_unshift( $this->curItem
, $elm );
965 * Start an element when in MODE_QDESC.
966 * This generally happens when a simple element has an inner
967 * rdf:Description to hold qualifier elements.
970 * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
971 * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
972 * </exif:DigitalZoomRatio>
973 * Called when processing the <rdf:value> or <foo:someQualifier>.
975 * @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
) {
1003 if ( isset( $this->items
[$ns][$tag] ) ) {
1004 if ( isset( $this->items
[$ns][$tag]['structPart'] ) ) {
1005 // If this element is supposed to appear only as
1006 // a child of a structure, but appears here (not as
1007 // a child of a struct), then something weird is
1008 // happening, so ignore this element and its children.
1010 $this->logger
->warning( "Encountered <$ns:$tag> outside"
1011 . " of its expected parent. Ignoring." );
1013 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1014 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1018 $mode = $this->items
[$ns][$tag]['mode'];
1019 array_unshift( $this->mode
, $mode );
1020 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1021 if ( $mode === self
::MODE_STRUCT
) {
1022 $this->ancestorStruct
= isset( $this->items
[$ns][$tag]['map_name'] )
1023 ?
$this->items
[$ns][$tag]['map_name'] : $tag;
1025 if ( $this->charContent
!== false ) {
1027 // Should not happen in valid XMP.
1028 throw new RuntimeException( 'tag nested in non-whitespace characters.' );
1031 // This element is not on our list of allowed elements so ignore.
1032 $this->logger
->debug( __METHOD__
. " Ignoring unrecognized element <$ns:$tag>." );
1033 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1034 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1039 // process attributes
1040 $this->doAttribs( $attribs );
1044 * Hit an opening element when in a Struct (MODE_STRUCT)
1045 * This is generally for fields of a compound property.
1047 * Example of a struct (abbreviated; flash has more properties):
1049 * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1050 * <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1054 * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1055 * <exif:Mode>1</exif:Mode></exif:Flash>
1057 * @param string $ns Namespace
1058 * @param string $tag Tag name (no ns)
1059 * @param array $attribs Array of attribs w/ values.
1060 * @throws RuntimeException
1062 private function startElementModeStruct( $ns, $tag, $attribs ) {
1063 if ( $ns !== self
::NS_RDF
) {
1065 if ( isset( $this->items
[$ns][$tag] ) ) {
1066 if ( isset( $this->items
[$ns][$this->ancestorStruct
]['children'] )
1067 && !isset( $this->items
[$ns][$this->ancestorStruct
]['children'][$tag] )
1069 // This assumes that we don't have inter-namespace nesting
1070 // which we don't in all the properties we're interested in.
1071 throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct
1072 . "> where it is not allowed." );
1074 array_unshift( $this->mode
, $this->items
[$ns][$tag]['mode'] );
1075 array_unshift( $this->curItem
, $ns . ' ' . $tag );
1076 if ( $this->charContent
!== false ) {
1078 // Should not happen in valid XMP.
1079 throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" .
1080 $this->charContent
. ")." );
1083 array_unshift( $this->mode
, self
::MODE_IGNORE
);
1084 array_unshift( $this->curItem
, $elm );
1090 if ( $ns === self
::NS_RDF
&& $tag === 'Description' ) {
1091 $this->doAttribs( $attribs );
1092 array_unshift( $this->mode
, self
::MODE_STRUCT
);
1093 array_unshift( $this->curItem
, $this->curItem
[0] );
1098 * opening element in MODE_LI
1099 * process elements of arrays.
1102 * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1103 * </rdf:Seq> </exif:ISOSpeedRatings>
1104 * This method is called when we hit the <rdf:li> element.
1106 * @param string $elm Namespace . ' ' . tagname
1107 * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1108 * @throws RuntimeException If gets a tag other than <rdf:li>
1110 private function startElementModeLi( $elm, $attribs ) {
1111 if ( ( $elm ) !== self
::NS_RDF
. ' li' ) {
1112 throw new RuntimeException( "<rdf:li> expected but got $elm." );
1115 if ( !isset( $this->mode
[1] ) ) {
1116 // This should never ever ever happen. Checking for it
1118 throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' );
1121 if ( $this->mode
[1] === self
::MODE_BAGSTRUCT
) {
1122 // This list item contains a compound (STRUCT) value.
1123 array_unshift( $this->mode
, self
::MODE_STRUCT
);
1124 array_unshift( $this->curItem
, $elm );
1125 $this->processingArray
= true;
1127 if ( !isset( $this->curItem
[1] ) ) {
1129 throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' );
1131 list( $curNS, $curTag ) = explode( ' ', $this->curItem
[1] );
1132 $this->ancestorStruct
= isset( $this->items
[$curNS][$curTag]['map_name'] )
1133 ?
$this->items
[$curNS][$curTag]['map_name'] : $curTag;
1135 $this->doAttribs( $attribs );
1137 // Normal BAG or SEQ containing simple values.
1138 array_unshift( $this->mode
, self
::MODE_SIMPLE
);
1139 // need to add curItem[0] on again since one is for the specific item
1140 // and one is for the entire group.
1141 array_unshift( $this->curItem
, $this->curItem
[0] );
1142 $this->processingArray
= true;
1147 * Opening element in MODE_LI_LANG.
1148 * process elements of language alternatives
1151 * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1152 * </rdf:li> </rdf:Alt> </dc:title>
1154 * This method is called when we hit the <rdf:li> element.
1156 * @param string $elm Namespace . ' ' . tag
1157 * @param array $attribs Array of elements (most importantly xml:lang)
1158 * @throws RuntimeException If gets a tag other than <rdf:li> or if no xml:lang
1160 private function startElementModeLiLang( $elm, $attribs ) {
1161 if ( $elm !== self
::NS_RDF
. ' li' ) {
1162 throw new RuntimeException( __METHOD__
. " <rdf:li> expected but got $elm." );
1164 if ( !isset( $attribs[self
::NS_XML
. ' lang'] )
1165 ||
!preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self
::NS_XML
. ' lang'] )
1167 throw new RuntimeException( __METHOD__
1168 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1171 // Lang is case-insensitive.
1172 $this->itemLang
= strtolower( $attribs[self
::NS_XML
. ' lang'] );
1174 // need to add curItem[0] on again since one is for the specific item
1175 // and one is for the entire group.
1176 array_unshift( $this->curItem
, $this->curItem
[0] );
1177 array_unshift( $this->mode
, self
::MODE_SIMPLE
);
1178 $this->processingArray
= true;
1182 * Hits an opening element.
1183 * Generally just calls a helper based on what MODE we're in.
1184 * Also does some initial set up for the wrapper element
1186 * @param XMLParser $parser
1187 * @param string $elm Namespace "<space>" element
1188 * @param array $attribs Attribute name => value
1189 * @throws RuntimeException
1191 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.
1210 // also it seems as if exiv2 and exiftool do not support
1211 // this either (That or I misunderstand the standard)
1212 $this->logger
->info( __METHOD__
. ' Encountered <rdf:type> which isn\'t currently supported' );
1215 if ( strpos( $elm, ' ' ) === false ) {
1216 // This probably shouldn't happen.
1217 $this->logger
->info( __METHOD__
. " Encountered <$elm> which has no namespace. Skipping." );
1222 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1224 if ( count( $this->mode
) === 0 ) {
1225 // This should not happen.
1226 throw new RuntimeException( 'Error extracting XMP, '
1227 . "encountered <$elm> with no mode" );
1230 switch ( $this->mode
[0] ) {
1231 case self
::MODE_IGNORE
:
1232 $this->startElementModeIgnore( $elm );
1234 case self
::MODE_SIMPLE
:
1235 $this->startElementModeSimple( $elm, $attribs );
1237 case self
::MODE_INITIAL
:
1238 $this->startElementModeInitial( $ns, $tag, $attribs );
1240 case self
::MODE_STRUCT
:
1241 $this->startElementModeStruct( $ns, $tag, $attribs );
1243 case self
::MODE_BAG
:
1244 case self
::MODE_BAGSTRUCT
:
1245 $this->startElementModeBag( $elm );
1247 case self
::MODE_SEQ
:
1248 $this->startElementModeSeq( $elm );
1250 case self
::MODE_LANG
:
1251 $this->startElementModeLang( $elm );
1253 case self
::MODE_LI_LANG
:
1254 $this->startElementModeLiLang( $elm, $attribs );
1257 $this->startElementModeLi( $elm, $attribs );
1259 case self
::MODE_QDESC
:
1260 $this->startElementModeQDesc( $elm );
1263 throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode
[0] );
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.
1274 * @codingStandardsIgnoreStart Long line that cannot be broken
1277 * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1279 * @codingStandardsIgnoreEnd
1281 * @param array $attribs Array attribute=>value
1282 * @throws RuntimeException
1284 private function doAttribs( $attribs ) {
1285 // first check for rdf:parseType attribute, as that can change
1286 // how the attributes are interperted.
1288 if ( isset( $attribs[self
::NS_RDF
. ' parseType'] )
1289 && $attribs[self
::NS_RDF
. ' parseType'] === 'Resource'
1290 && $this->mode
[0] === self
::MODE_SIMPLE
1292 // this is equivalent to having an inner rdf:Description
1293 $this->mode
[0] = self
::MODE_QDESC
;
1295 foreach ( $attribs as $name => $val ) {
1296 if ( strpos( $name, ' ' ) === false ) {
1297 // This shouldn't happen, but so far some old software forgets namespace
1299 $this->logger
->info( __METHOD__
. ' Encountered non-namespaced attribute: '
1300 . " $name=\"$val\". Skipping. " );
1303 list( $ns, $tag ) = explode( ' ', $name, 2 );
1304 if ( $ns === self
::NS_RDF
) {
1305 if ( $tag === 'value' ||
$tag === 'resource' ) {
1306 // resource is for url.
1307 // value attribute is a weird way of just putting the contents.
1308 $this->char( $this->xmlParser
, $val );
1310 } elseif ( isset( $this->items
[$ns][$tag] ) ) {
1311 if ( $this->mode
[0] === self
::MODE_SIMPLE
) {
1312 throw new RuntimeException( __METHOD__
1313 . " $ns:$tag found as attribute where not allowed" );
1315 $this->saveValue( $ns, $tag, $val );
1317 $this->logger
->debug( __METHOD__
. " Ignoring unrecognized element <$ns:$tag>." );
1323 * Given an extracted value, save it to results array
1325 * note also uses $this->ancestorStruct and
1326 * $this->processingArray to determine what name to
1327 * save the value under. (in addition to $tag).
1329 * @param string $ns Namespace of tag this is for
1330 * @param string $tag Tag name
1331 * @param string $val Value to save
1333 private function saveValue( $ns, $tag, $val ) {
1335 $info =& $this->items
[$ns][$tag];
1336 $finalName = isset( $info['map_name'] )
1337 ?
$info['map_name'] : $tag;
1338 if ( isset( $info['validate'] ) ) {
1339 if ( is_array( $info['validate'] ) ) {
1340 $validate = $info['validate'];
1342 $validator = new XMPValidate( $this->logger
);
1343 $validate = array( $validator, $info['validate'] );
1346 if ( is_callable( $validate ) ) {
1347 call_user_func_array( $validate, array( $info, &$val, true ) );
1348 // the reasoning behind using &$val instead of using the return value
1349 // is to be consistent between here and validating structures.
1350 if ( is_null( $val ) ) {
1351 $this->logger
->info( __METHOD__
. " <$ns:$tag> failed validation." );
1356 $this->logger
->warning( __METHOD__
. " Validation function for $finalName ("
1357 . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1361 if ( $this->ancestorStruct
&& $this->processingArray
) {
1362 // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1363 $this->results
['xmp-' . $info['map_group']][$this->ancestorStruct
][][$finalName] = $val;
1364 } elseif ( $this->ancestorStruct
) {
1365 $this->results
['xmp-' . $info['map_group']][$this->ancestorStruct
][$finalName] = $val;
1366 } elseif ( $this->processingArray
) {
1367 if ( $this->itemLang
=== false ) {
1369 $this->results
['xmp-' . $info['map_group']][$finalName][] = $val;
1372 $this->results
['xmp-' . $info['map_group']][$finalName][$this->itemLang
] = $val;
1375 $this->results
['xmp-' . $info['map_group']][$finalName] = $val;