includes/media/XMP.php

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