includes/media/XMP.php

   1 <?php
   2 /**
   3  * Reader for XMP data containing properties relevant to images.
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 2 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License along
  16  * with this program; if not, write to the Free Software Foundation, Inc.,
  17  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  18  * http://www.gnu.org/copyleft/gpl.html
  19  *
  20  * @file
  21  * @ingroup Media
  22  */
  23
  24 /**
  25  * Class for reading xmp data containing properties relevant to
  26  * images, and spitting out an array that FormatMetadata accepts.
  27  *
  28  * Note, this is not meant to recognize every possible thing you can
  29  * encode in XMP. It should recognize all the properties we want.
  30  * For example it doesn't have support for structures with multiple
  31  * nesting levels, as none of the properties we're supporting use that
  32  * feature. If it comes across properties it doesn't recognize, it should
  33  * ignore them.
  34  *
  35  * The public methods one would call in this class are
  36  * - parse( $content )
  37  *    Reads in xmp content.
  38  *    Can potentially be called multiple times with partial data each time.
  39  * - parseExtended( $content )
  40  *    Reads XMPExtended blocks (jpeg files only).
  41  * - getResults
  42  *    Outputs a results array.
  43  *
  44  * Note XMP kind of looks like rdf. They are not the same thing - XMP is
  45  * encoded as a specific subset of rdf. This class can read XMP. It cannot
  46  * read rdf.
  47  *
  48  */
  49 class XMPReader {
  50         /** @var array XMP item configuration array */
  51         protected $items;
  52
  53         /** @var array Array to hold the current element (and previous element, and so on) */
  54         private $curItem = array();
  55
  56         /** @var bool|string The structure name when processing nested structures. */
  57         private $ancestorStruct = false;
  58
  59         /** @var bool|string Temporary holder for character data that appears in xmp doc. */
  60         private $charContent = false;
  61
  62         /** @var array Stores the state the xmpreader is in (see MODE_FOO constants) */
  63         private $mode = array();
  64
  65         /** @var array Array to hold results */
  66         private $results = array();
  67
  68         /** @var bool If we're doing a seq or bag. */
  69         private $processingArray = false;
  70
  71         /** @var bool|string Used for lang alts only */
  72         private $itemLang = false;
  73
  74         /** @var resource A resource handle for the XML parser */
  75         private $xmlParser;
  76
  77         /** @var bool|string Character set like 'UTF-8' */
  78         private $charset = false;
  79
  80         /** @var int */
  81         private $extendedXMPOffset = 0;
  82
  83         /** @var int Flag determining if the XMP is safe to parse **/
  84         private $parsable = 0;
  85
  86         /** @var string Buffer of XML to parse **/
  87         private $xmlParsableBuffer = '';
  88
  89         /**
  90          * These are various mode constants.
  91          * they are used to figure out what to do
  92          * with an element when its encountered.
  93          *
  94          * For example, MODE_IGNORE is used when processing
  95          * a property we're not interested in. So if a new
  96          * element pops up when we're in that mode, we ignore it.
  97          */
  98         const MODE_INITIAL = 0;
  99         const MODE_IGNORE = 1;
 100         const MODE_LI = 2;
 101         const MODE_LI_LANG = 3;
 102         const MODE_QDESC = 4;
 103
 104         // The following MODE constants are also used in the
 105         // $items array to denote what type of property the item is.
 106         const MODE_SIMPLE = 10;
 107         const MODE_STRUCT = 11; // structure (associative array)
 108         const MODE_SEQ = 12; // ordered list
 109         const MODE_BAG = 13; // unordered list
 110         const MODE_LANG = 14;
 111         const MODE_ALT = 15; // non-language alt. Currently not implemented, and not needed atm.
 112         const MODE_BAGSTRUCT = 16; // A BAG of Structs.
 113
 114         const NS_RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#';
 115         const NS_XML = 'http://www.w3.org/XML/1998/namespace';
 116
 117         // States used while determining if XML is safe to parse
 118         const PARSABLE_UNKNOWN = 0;
 119         const PARSABLE_OK = 1;
 120         const PARSABLE_BUFFERING = 2;
 121         const PARSABLE_NO = 3;
 122
 123         /**
 124          * Constructor.
 125          *
 126          * Primary job is to initialize the XMLParser
 127          */
 128         function __construct() {
 129
 130                 if ( !function_exists( 'xml_parser_create_ns' ) ) {
 131                         // this should already be checked by this point
 132                         throw new MWException( 'XMP support requires XML Parser' );
 133                 }
 134
 135                 $this->items = XMPInfo::getItems();
 136
 137                 $this->resetXMLParser();
 138         }
 139
 140         /**
 141          * Main use is if a single item has multiple xmp documents describing it.
 142          * For example in jpeg's with extendedXMP
 143          */
 144         private function resetXMLParser() {
 145
 146                 if ( $this->xmlParser ) {
 147                         //is this needed?
 148                         xml_parser_free( $this->xmlParser );
 149                 }
 150
 151                 $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' );
 152                 xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 );
 153                 xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 );
 154
 155                 xml_set_element_handler( $this->xmlParser,
 156                         array( $this, 'startElement' ),
 157                         array( $this, 'endElement' ) );
 158
 159                 xml_set_character_data_handler( $this->xmlParser, array( $this, 'char' ) );
 160
 161                 $this->parsable = self::PARSABLE_UNKNOWN;
 162                 $this->xmlParsableBuffer = '';
 163         }
 164
 165         /** Destroy the xml parser
 166          *
 167          * Not sure if this is actually needed.
 168          */
 169         function __destruct() {
 170                 // not sure if this is needed.
 171                 xml_parser_free( $this->xmlParser );
 172         }
 173
 174         /**
 175          * Check if this instance supports using this class
 176          */
 177         public static function isSupported() {
 178                 return function_exists( 'xml_parser_create_ns' ) && class_exists( 'XMLReader' );
 179         }
 180
 181         /** Get the result array. Do some post-processing before returning
 182          * the array, and transform any metadata that is special-cased.
 183          *
 184          * @return array Array of results as an array of arrays suitable for
 185          *    FormatMetadata::getFormattedData().
 186          */
 187         public function getResults() {
 188                 // xmp-special is for metadata that affects how stuff
 189                 // is extracted. For example xmpNote:HasExtendedXMP.
 190
 191                 // It is also used to handle photoshop:AuthorsPosition
 192                 // which is weird and really part of another property,
 193                 // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard.
 194                 // The location fields also use it.
 195
 196                 $data = $this->results;
 197
 198                 Hooks::run( 'XMPGetResults', array( &$data ) );
 199
 200                 if ( isset( $data['xmp-special']['AuthorsPosition'] )
 201                         && is_string( $data['xmp-special']['AuthorsPosition'] )
 202                         && isset( $data['xmp-general']['Artist'][0] )
 203                 ) {
 204                         // Note, if there is more than one creator,
 205                         // this only applies to first. This also will
 206                         // only apply to the dc:Creator prop, not the
 207                         // exif:Artist prop.
 208
 209                         $data['xmp-general']['Artist'][0] =
 210                                 $data['xmp-special']['AuthorsPosition'] . ', '
 211                                 . $data['xmp-general']['Artist'][0];
 212                 }
 213
 214                 // Go through the LocationShown and LocationCreated
 215                 // changing it to the non-hierarchal form used by
 216                 // the other location fields.
 217
 218                 if ( isset( $data['xmp-special']['LocationShown'][0] )
 219                         && is_array( $data['xmp-special']['LocationShown'][0] )
 220                 ) {
 221                         // the is_array is just paranoia. It should always
 222                         // be an array.
 223                         foreach ( $data['xmp-special']['LocationShown'] as $loc ) {
 224                                 if ( !is_array( $loc ) ) {
 225                                         // To avoid copying over the _type meta-fields.
 226                                         continue;
 227                                 }
 228                                 foreach ( $loc as $field => $val ) {
 229                                         $data['xmp-general'][$field . 'Dest'][] = $val;
 230                                 }
 231                         }
 232                 }
 233                 if ( isset( $data['xmp-special']['LocationCreated'][0] )
 234                         && is_array( $data['xmp-special']['LocationCreated'][0] )
 235                 ) {
 236                         // the is_array is just paranoia. It should always
 237                         // be an array.
 238                         foreach ( $data['xmp-special']['LocationCreated'] as $loc ) {
 239                                 if ( !is_array( $loc ) ) {
 240                                         // To avoid copying over the _type meta-fields.
 241                                         continue;
 242                                 }
 243                                 foreach ( $loc as $field => $val ) {
 244                                         $data['xmp-general'][$field . 'Created'][] = $val;
 245                                 }
 246                         }
 247                 }
 248
 249                 // We don't want to return the special values, since they're
 250                 // special and not info to be stored about the file.
 251                 unset( $data['xmp-special'] );
 252
 253                 // Convert GPSAltitude to negative if below sea level.
 254                 if ( isset( $data['xmp-exif']['GPSAltitudeRef'] )
 255                         && isset( $data['xmp-exif']['GPSAltitude'] )
 256                 ) {
 257
 258                         // Must convert to a real before multiplying by -1
 259                         // XMPValidate guarantees there will always be a '/' in this value.
 260                         list( $nom, $denom ) = explode( '/', $data['xmp-exif']['GPSAltitude'] );
 261                         $data['xmp-exif']['GPSAltitude'] = $nom / $denom;
 262
 263                         if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) {
 264                                 $data['xmp-exif']['GPSAltitude'] *= -1;
 265                         }
 266                         unset( $data['xmp-exif']['GPSAltitudeRef'] );
 267                 }
 268
 269                 return $data;
 270         }
 271
 272         /**
 273          * Main function to call to parse XMP. Use getResults to
 274          * get results.
 275          *
 276          * Also catches any errors during processing, writes them to
 277          * debug log, blanks result array and returns false.
 278          *
 279          * @param string $content XMP data
 280          * @param bool $allOfIt If this is all the data (true) or if its split up (false). Default true
 281          * @param bool $reset Does xml parser need to be reset. Default false
 282          * @throws MWException
 283          * @return bool Success.
 284          */
 285         public function parse( $content, $allOfIt = true, $reset = false ) {
 286                 if ( $reset ) {
 287                         $this->resetXMLParser();
 288                 }
 289                 try {
 290
 291                         // detect encoding by looking for BOM which is supposed to be in processing instruction.
 292                         // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf
 293                         if ( !$this->charset ) {
 294                                 $bom = array();
 295                                 if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/',
 296                                         $content, $bom )
 297                                 ) {
 298                                         switch ( $bom[0] ) {
 299                                                 case "\xFE\xFF":
 300                                                         $this->charset = 'UTF-16BE';
 301                                                         break;
 302                                                 case "\xFF\xFE":
 303                                                         $this->charset = 'UTF-16LE';
 304                                                         break;
 305                                                 case "\x00\x00\xFE\xFF":
 306                                                         $this->charset = 'UTF-32BE';
 307                                                         break;
 308                                                 case "\xFF\xFE\x00\x00":
 309                                                         $this->charset = 'UTF-32LE';
 310                                                         break;
 311                                                 case "\xEF\xBB\xBF":
 312                                                         $this->charset = 'UTF-8';
 313                                                         break;
 314                                                 default:
 315                                                         //this should be impossible to get to
 316                                                         throw new MWException( "Invalid BOM" );
 317                                         }
 318                                 } else {
 319                                         // standard specifically says, if no bom assume utf-8
 320                                         $this->charset = 'UTF-8';
 321                                 }
 322                         }
 323                         if ( $this->charset !== 'UTF-8' ) {
 324                                 //don't convert if already utf-8
 325                                 wfSuppressWarnings();
 326                                 $content = iconv( $this->charset, 'UTF-8//IGNORE', $content );
 327                                 wfRestoreWarnings();
 328                         }
 329
 330                         // Ensure the XMP block does not have an xml doctype declaration, which
 331                         // could declare entities unsafe to parse with xml_parse (T85848/T71210).
 332                         if ( $this->parsable !== self::PARSABLE_OK ) {
 333                                 if ( $this->parsable === self::PARSABLE_NO ) {
 334                                         throw new Exception( 'Unsafe doctype declaration in XML.' );
 335                                 }
 336
 337                                 $content = $this->xmlParsableBuffer . $content;
 338                                 if ( !$this->checkParseSafety( $content ) ) {
 339                                         if ( !$allOfIt && $this->parsable !== self::PARSABLE_NO ) {
 340                                                 // parse wasn't Unsuccessful yet, so return true
 341                                                 // in this case.
 342                                                 return true;
 343                                         }
 344                                         $msg = ( $this->parsable === self::PARSABLE_NO ) ?
 345                                                 'Unsafe doctype declaration in XML.' :
 346                                                 'No root element found in XML.';
 347                                         throw new Exception( $msg );
 348                                 }
 349                         }
 350
 351                         $ok = xml_parse( $this->xmlParser, $content, $allOfIt );
 352                         if ( !$ok ) {
 353                                 $error = xml_error_string( xml_get_error_code( $this->xmlParser ) );
 354                                 $where = 'line: ' . xml_get_current_line_number( $this->xmlParser )
 355                                         . ' column: ' . xml_get_current_column_number( $this->xmlParser )
 356                                         . ' byte offset: ' . xml_get_current_byte_index( $this->xmlParser );
 357
 358                                 wfDebugLog( 'XMP', "XMPReader::parse : Error reading XMP content: $error ($where)" );
 359                                 $this->results = array(); // blank if error.
 360                                 return false;
 361                         }
 362                 } catch ( Exception $e ) {
 363                         wfDebugLog( 'XMP', 'XMP parse error: ' . $e );
 364                         $this->results = array();
 365
 366                         return false;
 367                 }
 368
 369                 return true;
 370         }
 371
 372         /** Entry point for XMPExtended blocks in jpeg files
 373          *
 374          * @todo In serious need of testing
 375          * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20
 376          * @param string $content XMPExtended block minus the namespace signature
 377          * @return bool If it succeeded.
 378          */
 379         public function parseExtended( $content ) {
 380                 // @todo FIXME: This is untested. Hard to find example files
 381                 // or programs that make such files..
 382                 $guid = substr( $content, 0, 32 );
 383                 if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] )
 384                         || $this->results['xmp-special']['HasExtendedXMP'] !== $guid
 385                 ) {
 386                         wfDebugLog( 'XMP', __METHOD__ .
 387                                 " Ignoring XMPExtended block due to wrong guid (guid= '$guid')" );
 388
 389                         return false;
 390                 }
 391                 $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) );
 392
 393                 if ( !$len || $len['length'] < 4 || $len['offset'] < 0 || $len['offset'] > $len['length'] ) {
 394                         wfDebugLog( 'XMP', __METHOD__ . 'Error reading extended XMP block, invalid length or offset.' );
 395
 396                         return false;
 397                 }
 398
 399                 // we're not very robust here. we should accept it in the wrong order.
 400                 // To quote the XMP standard:
 401                 // "A JPEG writer should write the ExtendedXMP marker segments in order,
 402                 // immediately following the StandardXMP. However, the JPEG standard
 403                 // does not require preservation of marker segment order. A robust JPEG
 404                 // reader should tolerate the marker segments in any order."
 405                 //
 406                 // otoh the probability that an image will have more than 128k of
 407                 // metadata is rather low... so the probability that it will have
 408                 // > 128k, and be in the wrong order is very low...
 409
 410                 if ( $len['offset'] !== $this->extendedXMPOffset ) {
 411                         wfDebugLog( 'XMP', __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was '
 412                                 . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')' );
 413
 414                         return false;
 415                 }
 416
 417                 if ( $len['offset'] === 0 ) {
 418                         // if we're starting the extended block, we've probably already
 419                         // done the XMPStandard block, so reset.
 420                         $this->resetXMLParser();
 421                 }
 422
 423                 $this->extendedXMPOffset += $len['length'];
 424
 425                 $actualContent = substr( $content, 40 );
 426
 427                 if ( $this->extendedXMPOffset === strlen( $actualContent ) ) {
 428                         $atEnd = true;
 429                 } else {
 430                         $atEnd = false;
 431                 }
 432
 433                 wfDebugLog( 'XMP', __METHOD__ . 'Parsing a XMPExtended block' );
 434
 435                 return $this->parse( $actualContent, $atEnd );
 436         }
 437
 438         /**
 439          * Character data handler
 440          * Called whenever character data is found in the xmp document.
 441          *
 442          * does nothing if we're in MODE_IGNORE or if the data is whitespace
 443          * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character
 444          * data in the other modes).
 445          *
 446          * As an example, this happens when we encounter XMP like:
 447          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 448          * and are processing the 0/10 bit.
 449          *
 450          * @param XMLParser $parser XMLParser reference to the xml parser
 451          * @param string $data Character data
 452          * @throws MWException On invalid data
 453          */
 454         function char( $parser, $data ) {
 455
 456                 $data = trim( $data );
 457                 if ( trim( $data ) === "" ) {
 458                         return;
 459                 }
 460
 461                 if ( !isset( $this->mode[0] ) ) {
 462                         throw new MWException( 'Unexpected character data before first rdf:Description element' );
 463                 }
 464
 465                 if ( $this->mode[0] === self::MODE_IGNORE ) {
 466                         return;
 467                 }
 468
 469                 if ( $this->mode[0] !== self::MODE_SIMPLE
 470                         && $this->mode[0] !== self::MODE_QDESC
 471                 ) {
 472                         throw new MWException( 'character data where not expected. (mode ' . $this->mode[0] . ')' );
 473                 }
 474
 475                 // to check, how does this handle w.s.
 476                 if ( $this->charContent === false ) {
 477                         $this->charContent = $data;
 478                 } else {
 479                         $this->charContent .= $data;
 480                 }
 481         }
 482
 483         /**
 484          * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't
 485          * contain a doctype declaration which could contain a dos attack if we
 486          * parse it and expand internal entities (T85848).
 487          *
 488          * @param string $content xml string to check for parse safety
 489          * @return bool true if the xml is safe to parse, false otherwise
 490          */
 491         private function checkParseSafety( $content ) {
 492                 $reader = new XMLReader();
 493                 $result = null;
 494
 495                 // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed
 496                 // instead of using XML().
 497                 $reader->open(
 498                         'data://text/plain,' . urlencode( $content ),
 499                         null,
 500                         LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET
 501                 );
 502
 503                 $oldDisable = libxml_disable_entity_loader( true );
 504                 $reset = new ScopedCallback(
 505                         'libxml_disable_entity_loader',
 506                         array( $oldDisable )
 507                 );
 508                 $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false );
 509
 510                 // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning
 511                 // when parsing truncated XML, which causes unit tests to fail.
 512                 wfSuppressWarnings();
 513                 while ( $reader->read() ) {
 514                         if ( $reader->nodeType === XMLReader::ELEMENT ) {
 515                                 // Reached the first element without hitting a doctype declaration
 516                                 $this->parsable = self::PARSABLE_OK;
 517                                 $result = true;
 518                                 break;
 519                         }
 520                         if ( $reader->nodeType === XMLReader::DOC_TYPE ) {
 521                                 $this->parsable = self::PARSABLE_NO;
 522                                 $result = false;
 523                                 break;
 524                         }
 525                 }
 526                 wfRestoreWarnings();
 527
 528                 if ( !is_null( $result ) ) {
 529                         return $result;
 530                 }
 531
 532                 // Reached the end of the parsable xml without finding an element
 533                 // or doctype. Buffer and try again.
 534                 $this->parsable = self::PARSABLE_BUFFERING;
 535                 $this->xmlParsableBuffer = $content;
 536                 return false;
 537         }
 538
 539         /** When we hit a closing element in MODE_IGNORE
 540          * Check to see if this is the element we started to ignore,
 541          * in which case we get out of MODE_IGNORE
 542          *
 543          * @param string $elm Namespace of element followed by a space and then tag name of element.
 544          */
 545         private function endElementModeIgnore( $elm ) {
 546                 if ( $this->curItem[0] === $elm ) {
 547                         array_shift( $this->curItem );
 548                         array_shift( $this->mode );
 549                 }
 550         }
 551
 552         /**
 553          * Hit a closing element when in MODE_SIMPLE.
 554          * This generally means that we finished processing a
 555          * property value, and now have to save the result to the
 556          * results array
 557          *
 558          * For example, when processing:
 559          * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio>
 560          * this deals with when we hit </exif:DigitalZoomRatio>.
 561          *
 562          * Or it could be if we hit the end element of a property
 563          * of a compound data structure (like a member of an array).
 564          *
 565          * @param string $elm Namespace, space, and tag name.
 566          */
 567         private function endElementModeSimple( $elm ) {
 568                 if ( $this->charContent !== false ) {
 569                         if ( $this->processingArray ) {
 570                                 // if we're processing an array, use the original element
 571                                 // name instead of rdf:li.
 572                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 573                         } else {
 574                                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 575                         }
 576                         $this->saveValue( $ns, $tag, $this->charContent );
 577
 578                         $this->charContent = false; // reset
 579                 }
 580                 array_shift( $this->curItem );
 581                 array_shift( $this->mode );
 582         }
 583
 584         /**
 585          * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG
 586          * generally means we've finished processing a nested structure.
 587          * resets some internal variables to indicate that.
 588          *
 589          * Note this means we hit the closing element not the "</rdf:Seq>".
 590          *
 591          * @par For example, when processing:
 592          * @code{,xml}
 593          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 594          *   </rdf:Seq> </exif:ISOSpeedRatings>
 595          * @endcode
 596          *
 597          * This method is called when we hit the "</exif:ISOSpeedRatings>" tag.
 598          *
 599          * @param string $elm Namespace . space . tag name.
 600          * @throws MWException
 601          */
 602         private function endElementNested( $elm ) {
 603
 604                 /* cur item must be the same as $elm, unless if in MODE_STRUCT
 605                    in which case it could also be rdf:Description */
 606                 if ( $this->curItem[0] !== $elm
 607                         && !( $elm === self::NS_RDF . ' Description'
 608                                 && $this->mode[0] === self::MODE_STRUCT )
 609                 ) {
 610                         throw new MWException( "nesting mismatch. got a </$elm> but expected a </" .
 611                                 $this->curItem[0] . '>' );
 612                 }
 613
 614                 // Validate structures.
 615                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
 616                 if ( isset( $this->items[$ns][$tag]['validate'] ) ) {
 617
 618                         $info =& $this->items[$ns][$tag];
 619                         $finalName = isset( $info['map_name'] )
 620                                 ? $info['map_name'] : $tag;
 621
 622                         $validate = is_array( $info['validate'] ) ? $info['validate']
 623                                 : array( 'XMPValidate', $info['validate'] );
 624
 625                         if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 626                                 // This can happen if all the members of the struct failed validation.
 627                                 wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> has no valid members." );
 628                         } elseif ( is_callable( $validate ) ) {
 629                                 $val =& $this->results['xmp-' . $info['map_group']][$finalName];
 630                                 call_user_func_array( $validate, array( $info, &$val, false ) );
 631                                 if ( is_null( $val ) ) {
 632                                         // the idea being the validation function will unset the variable if
 633                                         // its invalid.
 634                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
 635                                         unset( $this->results['xmp-' . $info['map_group']][$finalName] );
 636                                 }
 637                         } else {
 638                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
 639                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
 640                         }
 641                 }
 642
 643                 array_shift( $this->curItem );
 644                 array_shift( $this->mode );
 645                 $this->ancestorStruct = false;
 646                 $this->processingArray = false;
 647                 $this->itemLang = false;
 648         }
 649
 650         /**
 651          * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag )
 652          * Add information about what type of element this is.
 653          *
 654          * Note we still have to hit the outer "</property>"
 655          *
 656          * @par For example, when processing:
 657          * @code{,xml}
 658          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
 659          *   </rdf:Seq> </exif:ISOSpeedRatings>
 660          * @endcode
 661          *
 662          * This method is called when we hit the "</rdf:Seq>".
 663          * (For comparison, we call endElementModeSimple when we
 664          * hit the "</rdf:li>")
 665          *
 666          * @param string $elm Namespace . ' ' . element name
 667          * @throws MWException
 668          */
 669         private function endElementModeLi( $elm ) {
 670
 671                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 672                 $info = $this->items[$ns][$tag];
 673                 $finalName = isset( $info['map_name'] )
 674                         ? $info['map_name'] : $tag;
 675
 676                 array_shift( $this->mode );
 677
 678                 if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) {
 679                         wfDebugLog( 'XMP', __METHOD__ . " Empty compund element $finalName." );
 680
 681                         return;
 682                 }
 683
 684                 if ( $elm === self::NS_RDF . ' Seq' ) {
 685                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol';
 686                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 687                         $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul';
 688                 } elseif ( $elm === self::NS_RDF . ' Alt' ) {
 689                         // extra if needed as you could theoretically have a non-language alt.
 690                         if ( $info['mode'] === self::MODE_LANG ) {
 691                                 $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang';
 692                         }
 693                 } else {
 694                         throw new MWException( __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." );
 695                 }
 696         }
 697
 698         /**
 699          * End element while in MODE_QDESC
 700          * mostly when ending an element when we have a simple value
 701          * that has qualifiers.
 702          *
 703          * Qualifiers aren't all that common, and we don't do anything
 704          * with them.
 705          *
 706          * @param string $elm Namespace and element
 707          */
 708         private function endElementModeQDesc( $elm ) {
 709
 710                 if ( $elm === self::NS_RDF . ' value' ) {
 711                         list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 712                         $this->saveValue( $ns, $tag, $this->charContent );
 713
 714                         return;
 715                 } else {
 716                         array_shift( $this->mode );
 717                         array_shift( $this->curItem );
 718                 }
 719         }
 720
 721         /**
 722          * Handler for hitting a closing element.
 723          *
 724          * generally just calls a helper function depending on what
 725          * mode we're in.
 726          *
 727          * Ignores the outer wrapping elements that are optional in
 728          * xmp and have no meaning.
 729          *
 730          * @param XMLParser $parser
 731          * @param string $elm Namespace . ' ' . element name
 732          * @throws MWException
 733          */
 734         function endElement( $parser, $elm ) {
 735                 if ( $elm === ( self::NS_RDF . ' RDF' )
 736                         || $elm === 'adobe:ns:meta/ xmpmeta'
 737                         || $elm === 'adobe:ns:meta/ xapmeta'
 738                 ) {
 739                         // ignore these.
 740                         return;
 741                 }
 742
 743                 if ( $elm === self::NS_RDF . ' type' ) {
 744                         // these aren't really supported properly yet.
 745                         // However, it appears they almost never used.
 746                         wfDebugLog( 'XMP', __METHOD__ . ' encountered <rdf:type>' );
 747                 }
 748
 749                 if ( strpos( $elm, ' ' ) === false ) {
 750                         // This probably shouldn't happen.
 751                         // However, there is a bug in an adobe product
 752                         // that forgets the namespace on some things.
 753                         // (Luckily they are unimportant things).
 754                         wfDebugLog( 'XMP', __METHOD__ . " Encountered </$elm> which has no namespace. Skipping." );
 755
 756                         return;
 757                 }
 758
 759                 if ( count( $this->mode[0] ) === 0 ) {
 760                         // This should never ever happen and means
 761                         // there is a pretty major bug in this class.
 762                         throw new MWException( 'Encountered end element with no mode' );
 763                 }
 764
 765                 if ( count( $this->curItem ) == 0 && $this->mode[0] !== self::MODE_INITIAL ) {
 766                         // just to be paranoid. Should always have a curItem, except for initially
 767                         // (aka during MODE_INITAL).
 768                         throw new MWException( "Hit end element </$elm> but no curItem" );
 769                 }
 770
 771                 switch ( $this->mode[0] ) {
 772                         case self::MODE_IGNORE:
 773                                 $this->endElementModeIgnore( $elm );
 774                                 break;
 775                         case self::MODE_SIMPLE:
 776                                 $this->endElementModeSimple( $elm );
 777                                 break;
 778                         case self::MODE_STRUCT:
 779                         case self::MODE_SEQ:
 780                         case self::MODE_BAG:
 781                         case self::MODE_LANG:
 782                         case self::MODE_BAGSTRUCT:
 783                                 $this->endElementNested( $elm );
 784                                 break;
 785                         case self::MODE_INITIAL:
 786                                 if ( $elm === self::NS_RDF . ' Description' ) {
 787                                         array_shift( $this->mode );
 788                                 } else {
 789                                         throw new MWException( 'Element ended unexpectedly while in MODE_INITIAL' );
 790                                 }
 791                                 break;
 792                         case self::MODE_LI:
 793                         case self::MODE_LI_LANG:
 794                                 $this->endElementModeLi( $elm );
 795                                 break;
 796                         case self::MODE_QDESC:
 797                                 $this->endElementModeQDesc( $elm );
 798                                 break;
 799                         default:
 800                                 wfDebugLog( 'XMP', __METHOD__ . " no mode (elm = $elm)" );
 801                                 break;
 802                 }
 803         }
 804
 805         /**
 806          * Hit an opening element while in MODE_IGNORE
 807          *
 808          * XMP is extensible, so ignore any tag we don't understand.
 809          *
 810          * Mostly ignores, unless we encounter the element that we are ignoring.
 811          * in which case we add it to the item stack, so we can ignore things
 812          * that are nested, correctly.
 813          *
 814          * @param string $elm Namespace . ' ' . tag name
 815          */
 816         private function startElementModeIgnore( $elm ) {
 817                 if ( $elm === $this->curItem[0] ) {
 818                         array_unshift( $this->curItem, $elm );
 819                         array_unshift( $this->mode, self::MODE_IGNORE );
 820                 }
 821         }
 822
 823         /**
 824          *  Start element in MODE_BAG (unordered array)
 825          * this should always be <rdf:Bag>
 826          *
 827          * @param string $elm Namespace . ' ' . tag
 828          * @throws MWException If we have an element that's not <rdf:Bag>
 829          */
 830         private function startElementModeBag( $elm ) {
 831                 if ( $elm === self::NS_RDF . ' Bag' ) {
 832                         array_unshift( $this->mode, self::MODE_LI );
 833                 } else {
 834                         throw new MWException( "Expected <rdf:Bag> but got $elm." );
 835                 }
 836         }
 837
 838         /**
 839          * Start element in MODE_SEQ (ordered array)
 840          * this should always be <rdf:Seq>
 841          *
 842          * @param string $elm Namespace . ' ' . tag
 843          * @throws MWException If we have an element that's not <rdf:Seq>
 844          */
 845         private function startElementModeSeq( $elm ) {
 846                 if ( $elm === self::NS_RDF . ' Seq' ) {
 847                         array_unshift( $this->mode, self::MODE_LI );
 848                 } elseif ( $elm === self::NS_RDF . ' Bag' ) {
 849                         # bug 27105
 850                         wfDebugLog( 'XMP', __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending'
 851                                 . ' it is a Seq, since some buggy software is known to screw this up.' );
 852                         array_unshift( $this->mode, self::MODE_LI );
 853                 } else {
 854                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 855                 }
 856         }
 857
 858         /**
 859          * Start element in MODE_LANG (language alternative)
 860          * this should always be <rdf:Alt>
 861          *
 862          * This tag tends to be used for metadata like describe this
 863          * picture, which can be translated into multiple languages.
 864          *
 865          * XMP supports non-linguistic alternative selections,
 866          * which are really only used for thumbnails, which
 867          * we don't care about.
 868          *
 869          * @param string $elm Namespace . ' ' . tag
 870          * @throws MWException If we have an element that's not <rdf:Alt>
 871          */
 872         private function startElementModeLang( $elm ) {
 873                 if ( $elm === self::NS_RDF . ' Alt' ) {
 874                         array_unshift( $this->mode, self::MODE_LI_LANG );
 875                 } else {
 876                         throw new MWException( "Expected <rdf:Seq> but got $elm." );
 877                 }
 878         }
 879
 880         /**
 881          * Handle an opening element when in MODE_SIMPLE
 882          *
 883          * This should not happen often. This is for if a simple element
 884          * already opened has a child element. Could happen for a
 885          * qualified element.
 886          *
 887          * For example:
 888          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 889          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 890          *   </exif:DigitalZoomRatio>
 891          *
 892          * This method is called when processing the <rdf:Description> element
 893          *
 894          * @param string $elm Namespace and tag names separated by space.
 895          * @param array $attribs Attributes of the element.
 896          * @throws MWException
 897          */
 898         private function startElementModeSimple( $elm, $attribs ) {
 899                 if ( $elm === self::NS_RDF . ' Description' ) {
 900                         // If this value has qualifiers
 901                         array_unshift( $this->mode, self::MODE_QDESC );
 902                         array_unshift( $this->curItem, $this->curItem[0] );
 903
 904                         if ( isset( $attribs[self::NS_RDF . ' value'] ) ) {
 905                                 list( $ns, $tag ) = explode( ' ', $this->curItem[0], 2 );
 906                                 $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] );
 907                         }
 908                 } elseif ( $elm === self::NS_RDF . ' value' ) {
 909                         // This should not be here.
 910                         throw new MWException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' );
 911                 } else {
 912                         // something else we don't recognize, like a qualifier maybe.
 913                         wfDebugLog( 'XMP', __METHOD__ .
 914                                 " Encountered element <$elm> where only expecting character data as value of " .
 915                                 $this->curItem[0] );
 916                         array_unshift( $this->mode, self::MODE_IGNORE );
 917                         array_unshift( $this->curItem, $elm );
 918                 }
 919         }
 920
 921         /**
 922          * Start an element when in MODE_QDESC.
 923          * This generally happens when a simple element has an inner
 924          * rdf:Description to hold qualifier elements.
 925          *
 926          * For example in:
 927          * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value>
 928          *   <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description>
 929          *   </exif:DigitalZoomRatio>
 930          * Called when processing the <rdf:value> or <foo:someQualifier>.
 931          *
 932          * @param string $elm Namespace and tag name separated by a space.
 933          *
 934          */
 935         private function startElementModeQDesc( $elm ) {
 936                 if ( $elm === self::NS_RDF . ' value' ) {
 937                         return; // do nothing
 938                 } else {
 939                         // otherwise its a qualifier, which we ignore
 940                         array_unshift( $this->mode, self::MODE_IGNORE );
 941                         array_unshift( $this->curItem, $elm );
 942                 }
 943         }
 944
 945         /**
 946          * Starting an element when in MODE_INITIAL
 947          * This usually happens when we hit an element inside
 948          * the outer rdf:Description
 949          *
 950          * This is generally where most properties start.
 951          *
 952          * @param string $ns Namespace
 953          * @param string $tag Tag name (without namespace prefix)
 954          * @param array $attribs Array of attributes
 955          * @throws MWException
 956          */
 957         private function startElementModeInitial( $ns, $tag, $attribs ) {
 958                 if ( $ns !== self::NS_RDF ) {
 959
 960                         if ( isset( $this->items[$ns][$tag] ) ) {
 961                                 if ( isset( $this->items[$ns][$tag]['structPart'] ) ) {
 962                                         // If this element is supposed to appear only as
 963                                         // a child of a structure, but appears here (not as
 964                                         // a child of a struct), then something weird is
 965                                         // happening, so ignore this element and its children.
 966
 967                                         wfDebugLog( 'XMP', "Encountered <$ns:$tag> outside"
 968                                                 . " of its expected parent. Ignoring." );
 969
 970                                         array_unshift( $this->mode, self::MODE_IGNORE );
 971                                         array_unshift( $this->curItem, $ns . ' ' . $tag );
 972
 973                                         return;
 974                                 }
 975                                 $mode = $this->items[$ns][$tag]['mode'];
 976                                 array_unshift( $this->mode, $mode );
 977                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 978                                 if ( $mode === self::MODE_STRUCT ) {
 979                                         $this->ancestorStruct = isset( $this->items[$ns][$tag]['map_name'] )
 980                                                 ? $this->items[$ns][$tag]['map_name'] : $tag;
 981                                 }
 982                                 if ( $this->charContent !== false ) {
 983                                         // Something weird.
 984                                         // Should not happen in valid XMP.
 985                                         throw new MWException( 'tag nested in non-whitespace characters.' );
 986                                 }
 987                         } else {
 988                                 // This element is not on our list of allowed elements so ignore.
 989                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
 990                                 array_unshift( $this->mode, self::MODE_IGNORE );
 991                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
 992
 993                                 return;
 994                         }
 995                 }
 996                 // process attributes
 997                 $this->doAttribs( $attribs );
 998         }
 999
1000         /**
1001          * Hit an opening element when in a Struct (MODE_STRUCT)
1002          * This is generally for fields of a compound property.
1003          *
1004          * Example of a struct (abbreviated; flash has more properties):
1005          *
1006          * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired>
1007          *  <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash>
1008          *
1009          * or:
1010          *
1011          * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired>
1012          *  <exif:Mode>1</exif:Mode></exif:Flash>
1013          *
1014          * @param string $ns Namespace
1015          * @param string $tag Tag name (no ns)
1016          * @param array $attribs Array of attribs w/ values.
1017          * @throws MWException
1018          */
1019         private function startElementModeStruct( $ns, $tag, $attribs ) {
1020                 if ( $ns !== self::NS_RDF ) {
1021
1022                         if ( isset( $this->items[$ns][$tag] ) ) {
1023                                 if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] )
1024                                         && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] )
1025                                 ) {
1026                                         // This assumes that we don't have inter-namespace nesting
1027                                         // which we don't in all the properties we're interested in.
1028                                         throw new MWException( " <$tag> appeared nested in <" . $this->ancestorStruct
1029                                                 . "> where it is not allowed." );
1030                                 }
1031                                 array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] );
1032                                 array_unshift( $this->curItem, $ns . ' ' . $tag );
1033                                 if ( $this->charContent !== false ) {
1034                                         // Something weird.
1035                                         // Should not happen in valid XMP.
1036                                         throw new MWException( "tag <$tag> nested in non-whitespace characters (" .
1037                                                 $this->charContent . ")." );
1038                                 }
1039                         } else {
1040                                 array_unshift( $this->mode, self::MODE_IGNORE );
1041                                 array_unshift( $this->curItem, $elm );
1042
1043                                 return;
1044                         }
1045                 }
1046
1047                 if ( $ns === self::NS_RDF && $tag === 'Description' ) {
1048                         $this->doAttribs( $attribs );
1049                         array_unshift( $this->mode, self::MODE_STRUCT );
1050                         array_unshift( $this->curItem, $this->curItem[0] );
1051                 }
1052         }
1053
1054         /**
1055          * opening element in MODE_LI
1056          * process elements of arrays.
1057          *
1058          * Example:
1059          * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li>
1060          *   </rdf:Seq> </exif:ISOSpeedRatings>
1061          * This method is called when we hit the <rdf:li> element.
1062          *
1063          * @param string $elm Namespace . ' ' . tagname
1064          * @param array $attribs Attributes. (needed for BAGSTRUCTS)
1065          * @throws MWException If gets a tag other than <rdf:li>
1066          */
1067         private function startElementModeLi( $elm, $attribs ) {
1068                 if ( ( $elm ) !== self::NS_RDF . ' li' ) {
1069                         throw new MWException( "<rdf:li> expected but got $elm." );
1070                 }
1071
1072                 if ( !isset( $this->mode[1] ) ) {
1073                         // This should never ever ever happen. Checking for it
1074                         // to be paranoid.
1075                         throw new MWException( 'In mode Li, but no 2xPrevious mode!' );
1076                 }
1077
1078                 if ( $this->mode[1] === self::MODE_BAGSTRUCT ) {
1079                         // This list item contains a compound (STRUCT) value.
1080                         array_unshift( $this->mode, self::MODE_STRUCT );
1081                         array_unshift( $this->curItem, $elm );
1082                         $this->processingArray = true;
1083
1084                         if ( !isset( $this->curItem[1] ) ) {
1085                                 // be paranoid.
1086                                 throw new MWException( 'Can not find parent of BAGSTRUCT.' );
1087                         }
1088                         list( $curNS, $curTag ) = explode( ' ', $this->curItem[1] );
1089                         $this->ancestorStruct = isset( $this->items[$curNS][$curTag]['map_name'] )
1090                                 ? $this->items[$curNS][$curTag]['map_name'] : $curTag;
1091
1092                         $this->doAttribs( $attribs );
1093                 } else {
1094                         // Normal BAG or SEQ containing simple values.
1095                         array_unshift( $this->mode, self::MODE_SIMPLE );
1096                         // need to add curItem[0] on again since one is for the specific item
1097                         // and one is for the entire group.
1098                         array_unshift( $this->curItem, $this->curItem[0] );
1099                         $this->processingArray = true;
1100                 }
1101         }
1102
1103         /**
1104          * Opening element in MODE_LI_LANG.
1105          * process elements of language alternatives
1106          *
1107          * Example:
1108          * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house
1109          *  </rdf:li> </rdf:Alt> </dc:title>
1110          *
1111          * This method is called when we hit the <rdf:li> element.
1112          *
1113          * @param string $elm Namespace . ' ' . tag
1114          * @param array $attribs Array of elements (most importantly xml:lang)
1115          * @throws MWException If gets a tag other than <rdf:li> or if no xml:lang
1116          */
1117         private function startElementModeLiLang( $elm, $attribs ) {
1118                 if ( $elm !== self::NS_RDF . ' li' ) {
1119                         throw new MWException( __METHOD__ . " <rdf:li> expected but got $elm." );
1120                 }
1121                 if ( !isset( $attribs[self::NS_XML . ' lang'] )
1122                         || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] )
1123                 ) {
1124                         throw new MWException( __METHOD__
1125                                 . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" );
1126                 }
1127
1128                 // Lang is case-insensitive.
1129                 $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] );
1130
1131                 // need to add curItem[0] on again since one is for the specific item
1132                 // and one is for the entire group.
1133                 array_unshift( $this->curItem, $this->curItem[0] );
1134                 array_unshift( $this->mode, self::MODE_SIMPLE );
1135                 $this->processingArray = true;
1136         }
1137
1138         /**
1139          * Hits an opening element.
1140          * Generally just calls a helper based on what MODE we're in.
1141          * Also does some initial set up for the wrapper element
1142          *
1143          * @param XMLParser $parser
1144          * @param string $elm Namespace "<space>" element
1145          * @param array $attribs Attribute name => value
1146          * @throws MWException
1147          */
1148         function startElement( $parser, $elm, $attribs ) {
1149
1150                 if ( $elm === self::NS_RDF . ' RDF'
1151                         || $elm === 'adobe:ns:meta/ xmpmeta'
1152                         || $elm === 'adobe:ns:meta/ xapmeta'
1153                 ) {
1154                         /* ignore. */
1155                         return;
1156                 } elseif ( $elm === self::NS_RDF . ' Description' ) {
1157                         if ( count( $this->mode ) === 0 ) {
1158                                 // outer rdf:desc
1159                                 array_unshift( $this->mode, self::MODE_INITIAL );
1160                         }
1161                 } elseif ( $elm === self::NS_RDF . ' type' ) {
1162                         // This doesn't support rdf:type properly.
1163                         // In practise I have yet to see a file that
1164                         // uses this element, however it is mentioned
1165                         // on page 25 of part 1 of the xmp standard.
1166                         //
1167                         // also it seems as if exiv2 and exiftool do not support
1168                         // this either (That or I misunderstand the standard)
1169                         wfDebugLog( 'XMP', __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported' );
1170                 }
1171
1172                 if ( strpos( $elm, ' ' ) === false ) {
1173                         // This probably shouldn't happen.
1174                         wfDebugLog( 'XMP', __METHOD__ . " Encountered <$elm> which has no namespace. Skipping." );
1175
1176                         return;
1177                 }
1178
1179                 list( $ns, $tag ) = explode( ' ', $elm, 2 );
1180
1181                 if ( count( $this->mode ) === 0 ) {
1182                         // This should not happen.
1183                         throw new MWException( 'Error extracting XMP, '
1184                                 . "encountered <$elm> with no mode" );
1185                 }
1186
1187                 switch ( $this->mode[0] ) {
1188                         case self::MODE_IGNORE:
1189                                 $this->startElementModeIgnore( $elm );
1190                                 break;
1191                         case self::MODE_SIMPLE:
1192                                 $this->startElementModeSimple( $elm, $attribs );
1193                                 break;
1194                         case self::MODE_INITIAL:
1195                                 $this->startElementModeInitial( $ns, $tag, $attribs );
1196                                 break;
1197                         case self::MODE_STRUCT:
1198                                 $this->startElementModeStruct( $ns, $tag, $attribs );
1199                                 break;
1200                         case self::MODE_BAG:
1201                         case self::MODE_BAGSTRUCT:
1202                                 $this->startElementModeBag( $elm );
1203                                 break;
1204                         case self::MODE_SEQ:
1205                                 $this->startElementModeSeq( $elm );
1206                                 break;
1207                         case self::MODE_LANG:
1208                                 $this->startElementModeLang( $elm );
1209                                 break;
1210                         case self::MODE_LI_LANG:
1211                                 $this->startElementModeLiLang( $elm, $attribs );
1212                                 break;
1213                         case self::MODE_LI:
1214                                 $this->startElementModeLi( $elm, $attribs );
1215                                 break;
1216                         case self::MODE_QDESC:
1217                                 $this->startElementModeQDesc( $elm );
1218                                 break;
1219                         default:
1220                                 throw new MWException( 'StartElement in unknown mode: ' . $this->mode[0] );
1221                 }
1222         }
1223
1224         /**
1225          * Process attributes.
1226          * Simple values can be stored as either a tag or attribute
1227          *
1228          * Often the initial "<rdf:Description>" tag just has all the simple
1229          * properties as attributes.
1230          *
1231          * @codingStandardsIgnoreStart Long line that cannot be broken
1232          * @par Example:
1233          * @code
1234          * <rdf:Description rdf:about="" xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10">
1235          * @endcode
1236          * @codingStandardsIgnoreEnd
1237          *
1238          * @param array $attribs Array attribute=>value
1239          * @throws MWException
1240          */
1241         private function doAttribs( $attribs ) {
1242                 // first check for rdf:parseType attribute, as that can change
1243                 // how the attributes are interperted.
1244
1245                 if ( isset( $attribs[self::NS_RDF . ' parseType'] )
1246                         && $attribs[self::NS_RDF . ' parseType'] === 'Resource'
1247                         && $this->mode[0] === self::MODE_SIMPLE
1248                 ) {
1249                         // this is equivalent to having an inner rdf:Description
1250                         $this->mode[0] = self::MODE_QDESC;
1251                 }
1252                 foreach ( $attribs as $name => $val ) {
1253                         if ( strpos( $name, ' ' ) === false ) {
1254                                 // This shouldn't happen, but so far some old software forgets namespace
1255                                 // on rdf:about.
1256                                 wfDebugLog( 'XMP', __METHOD__ . ' Encountered non-namespaced attribute: '
1257                                         . " $name=\"$val\". Skipping. " );
1258                                 continue;
1259                         }
1260                         list( $ns, $tag ) = explode( ' ', $name, 2 );
1261                         if ( $ns === self::NS_RDF ) {
1262                                 if ( $tag === 'value' || $tag === 'resource' ) {
1263                                         // resource is for url.
1264                                         // value attribute is a weird way of just putting the contents.
1265                                         $this->char( $this->xmlParser, $val );
1266                                 }
1267                         } elseif ( isset( $this->items[$ns][$tag] ) ) {
1268                                 if ( $this->mode[0] === self::MODE_SIMPLE ) {
1269                                         throw new MWException( __METHOD__
1270                                                 . " $ns:$tag found as attribute where not allowed" );
1271                                 }
1272                                 $this->saveValue( $ns, $tag, $val );
1273                         } else {
1274                                 wfDebugLog( 'XMP', __METHOD__ . " Ignoring unrecognized element <$ns:$tag>." );
1275                         }
1276                 }
1277         }
1278
1279         /**
1280          * Given an extracted value, save it to results array
1281          *
1282          * note also uses $this->ancestorStruct and
1283          * $this->processingArray to determine what name to
1284          * save the value under. (in addition to $tag).
1285          *
1286          * @param string $ns Namespace of tag this is for
1287          * @param string $tag Tag name
1288          * @param string $val Value to save
1289          */
1290         private function saveValue( $ns, $tag, $val ) {
1291
1292                 $info =& $this->items[$ns][$tag];
1293                 $finalName = isset( $info['map_name'] )
1294                         ? $info['map_name'] : $tag;
1295                 if ( isset( $info['validate'] ) ) {
1296                         $validate = is_array( $info['validate'] ) ? $info['validate']
1297                                 : array( 'XMPValidate', $info['validate'] );
1298
1299                         if ( is_callable( $validate ) ) {
1300                                 call_user_func_array( $validate, array( $info, &$val, true ) );
1301                                 // the reasoning behind using &$val instead of using the return value
1302                                 // is to be consistent between here and validating structures.
1303                                 if ( is_null( $val ) ) {
1304                                         wfDebugLog( 'XMP', __METHOD__ . " <$ns:$tag> failed validation." );
1305
1306                                         return;
1307                                 }
1308                         } else {
1309                                 wfDebugLog( 'XMP', __METHOD__ . " Validation function for $finalName ("
1310                                         . $validate[0] . '::' . $validate[1] . '()) is not callable.' );
1311                         }
1312                 }
1313
1314                 if ( $this->ancestorStruct && $this->processingArray ) {
1315                         // Aka both an array and a struct. ( self::MODE_BAGSTRUCT )
1316                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val;
1317                 } elseif ( $this->ancestorStruct ) {
1318                         $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val;
1319                 } elseif ( $this->processingArray ) {
1320                         if ( $this->itemLang === false ) {
1321                                 // normal array
1322                                 $this->results['xmp-' . $info['map_group']][$finalName][] = $val;
1323                         } else {
1324                                 // lang array.
1325                                 $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val;
1326                         }
1327                 } else {
1328                         $this->results['xmp-' . $info['map_group']][$finalName] = $val;
1329                 }
1330         }
1331 }