Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
71.89% |
381 / 530 |
|
28.57% |
8 / 28 |
CRAP | |
0.00% |
0 / 1 |
| Reader | |
71.89% |
381 / 530 |
|
28.57% |
8 / 28 |
1011.09 | |
0.00% |
0 / 1 |
| __construct | |
83.33% |
5 / 6 |
|
0.00% |
0 / 1 |
2.02 | |||
| destroyXMLParser | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| resetXMLParser | |
100.00% |
10 / 10 |
|
100.00% |
1 / 1 |
1 | |||
| isSupported | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
2 | |||
| getResults | |
56.67% |
17 / 30 |
|
0.00% |
0 / 1 |
40.52 | |||
| parse | |
69.62% |
55 / 79 |
|
0.00% |
0 / 1 |
31.22 | |||
| parseExtended | |
86.11% |
31 / 36 |
|
0.00% |
0 / 1 |
9.22 | |||
| char | |
76.92% |
10 / 13 |
|
0.00% |
0 / 1 |
7.60 | |||
| checkParseSafety | |
78.79% |
26 / 33 |
|
0.00% |
0 / 1 |
8.61 | |||
| endElementModeIgnore | |
100.00% |
3 / 3 |
|
100.00% |
1 / 1 |
2 | |||
| endElementModeSimple | |
100.00% |
8 / 8 |
|
100.00% |
1 / 1 |
3 | |||
| endElementNested | |
67.57% |
25 / 37 |
|
0.00% |
0 / 1 |
11.76 | |||
| endElementModeLi | |
55.00% |
11 / 20 |
|
0.00% |
0 / 1 |
9.28 | |||
| endElementModeQDesc | |
100.00% |
6 / 6 |
|
100.00% |
1 / 1 |
2 | |||
| endElement | |
56.41% |
22 / 39 |
|
0.00% |
0 / 1 |
62.09 | |||
| startElementModeIgnore | |
33.33% |
1 / 3 |
|
0.00% |
0 / 1 |
3.19 | |||
| startElementModeBag | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| startElementModeSeq | |
80.00% |
8 / 10 |
|
0.00% |
0 / 1 |
3.07 | |||
| startElementModeLang | |
66.67% |
2 / 3 |
|
0.00% |
0 / 1 |
2.15 | |||
| startElementModeSimple | |
33.33% |
6 / 18 |
|
0.00% |
0 / 1 |
8.74 | |||
| startElementModeQDesc | |
100.00% |
4 / 4 |
|
100.00% |
1 / 1 |
2 | |||
| startElementModeInitial | |
100.00% |
23 / 23 |
|
100.00% |
1 / 1 |
6 | |||
| startElementModeStruct | |
72.22% |
13 / 18 |
|
0.00% |
0 / 1 |
9.37 | |||
| startElementModeLi | |
37.50% |
6 / 16 |
|
0.00% |
0 / 1 |
11.10 | |||
| startElementModeLiLang | |
70.00% |
7 / 10 |
|
0.00% |
0 / 1 |
4.43 | |||
| startElement | |
72.73% |
32 / 44 |
|
0.00% |
0 / 1 |
29.95 | |||
| doAttribs | |
88.00% |
22 / 25 |
|
0.00% |
0 / 1 |
11.21 | |||
| saveValue | |
75.86% |
22 / 29 |
|
0.00% |
0 / 1 |
11.41 | |||
| 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 | namespace Wikimedia\XMPReader; |
| 25 | |
| 26 | use Exception; |
| 27 | use Psr\Log\LoggerAwareInterface; |
| 28 | use Psr\Log\LoggerAwareTrait; |
| 29 | use Psr\Log\LoggerInterface; |
| 30 | use Psr\Log\NullLogger; |
| 31 | use RuntimeException; |
| 32 | use Wikimedia\ScopedCallback; |
| 33 | use XMLReader; |
| 34 | |
| 35 | /** |
| 36 | * Class for reading xmp data containing properties relevant to |
| 37 | * images, and spitting out an array that FormatMetadata accepts. |
| 38 | * |
| 39 | * Note, this is not meant to recognize every possible thing you can |
| 40 | * encode in XMP. It should recognize all the properties we want. |
| 41 | * For example, it doesn't have support for structures with multiple |
| 42 | * nesting levels, as none of the properties we're supporting use that |
| 43 | * feature. If it comes across properties it doesn't recognize, it should |
| 44 | * ignore them. |
| 45 | * |
| 46 | * The public methods one would call in this class are |
| 47 | * - parse( $content ) |
| 48 | * Reads in xmp content. |
| 49 | * Can potentially be called multiple times with partial data each time. |
| 50 | * - parseExtended( $content ) |
| 51 | * Reads XMPExtended blocks (jpeg files only). |
| 52 | * - getResults |
| 53 | * Outputs a results array. |
| 54 | * |
| 55 | * Note XMP kind of looks like rdf. They are not the same thing - XMP is |
| 56 | * encoded as a specific subset of rdf. This class can read XMP. It cannot |
| 57 | * read rdf. |
| 58 | */ |
| 59 | class Reader implements LoggerAwareInterface { |
| 60 | use LoggerAwareTrait; |
| 61 | |
| 62 | /** @var array XMP item configuration array */ |
| 63 | protected array $items; |
| 64 | |
| 65 | /** @var array Array to hold the current element (and previous element, and so on) */ |
| 66 | private array $curItem = []; |
| 67 | |
| 68 | /** @var bool|string The structure name when processing nested structures. */ |
| 69 | private $ancestorStruct = false; |
| 70 | |
| 71 | /** @var bool|string Temporary holder for character data that appears in xmp doc. */ |
| 72 | private $charContent = false; |
| 73 | |
| 74 | /** @var array Stores the state the xmpreader is in (see MODE_FOO constants) */ |
| 75 | private array $mode = []; |
| 76 | |
| 77 | /** @var array Array to hold results */ |
| 78 | private array $results = []; |
| 79 | |
| 80 | /** @var bool If we're doing a seq or bag. */ |
| 81 | private bool $processingArray = false; |
| 82 | |
| 83 | /** @var bool|string Used for lang alts only */ |
| 84 | private $itemLang = false; |
| 85 | |
| 86 | /** @var resource|null A resource handle for the XML parser */ |
| 87 | private $xmlParser; |
| 88 | |
| 89 | /** @var bool|string Character set like 'UTF-8' */ |
| 90 | private $charset = false; |
| 91 | |
| 92 | /** @var int */ |
| 93 | private int $extendedXMPOffset = 0; |
| 94 | |
| 95 | /** @var int Flag determining if the XMP is safe to parse */ |
| 96 | private int $parsable = 0; |
| 97 | |
| 98 | /** @var string Buffer of XML to parse */ |
| 99 | private string $xmlParsableBuffer = ''; |
| 100 | |
| 101 | /** |
| 102 | * @var string |
| 103 | */ |
| 104 | private string $filename; |
| 105 | |
| 106 | /** |
| 107 | * These are various mode constants. |
| 108 | * they are used to figure out what to do |
| 109 | * with an element when its encountered. |
| 110 | * |
| 111 | * For example, MODE_IGNORE is used when processing |
| 112 | * a property we're not interested in. So if a new |
| 113 | * element pops up when we're in that mode, we ignore it. |
| 114 | */ |
| 115 | private const MODE_INITIAL = 0; |
| 116 | private const MODE_IGNORE = 1; |
| 117 | private const MODE_LI = 2; |
| 118 | private const MODE_LI_LANG = 3; |
| 119 | private const MODE_QDESC = 4; |
| 120 | |
| 121 | // The following MODE constants are also used in the |
| 122 | // $items array to denote what type of property the item is. |
| 123 | public const MODE_SIMPLE = 10; |
| 124 | // structure (associative array) |
| 125 | public const MODE_STRUCT = 11; |
| 126 | // ordered list |
| 127 | public const MODE_SEQ = 12; |
| 128 | // unordered list |
| 129 | public const MODE_BAG = 13; |
| 130 | public const MODE_LANG = 14; |
| 131 | // non-language alt. Currently not implemented, and not needed atm. |
| 132 | public const MODE_ALT = 15; |
| 133 | // A BAG of Structs. |
| 134 | public const MODE_BAGSTRUCT = 16; |
| 135 | |
| 136 | private const NS_RDF = 'http://www.w3.org/1999/02/22-rdf-syntax-ns#'; |
| 137 | private const NS_XML = 'http://www.w3.org/XML/1998/namespace'; |
| 138 | |
| 139 | // States used while determining if XML is safe to parse |
| 140 | private const PARSABLE_UNKNOWN = 0; |
| 141 | private const PARSABLE_OK = 1; |
| 142 | private const PARSABLE_BUFFERING = 2; |
| 143 | private const PARSABLE_NO = 3; |
| 144 | |
| 145 | /** |
| 146 | * Primary job is to initialize the XMLParser |
| 147 | * |
| 148 | * @param LoggerInterface|null $logger Logger instance if available |
| 149 | * @param string $filename |
| 150 | */ |
| 151 | public function __construct( ?LoggerInterface $logger = null, $filename = 'unknown' ) { |
| 152 | if ( $logger ) { |
| 153 | $this->setLogger( $logger ); |
| 154 | } else { |
| 155 | $this->setLogger( new NullLogger() ); |
| 156 | } |
| 157 | $this->filename = $filename; |
| 158 | |
| 159 | $this->items = Info::getItems(); |
| 160 | |
| 161 | $this->resetXMLParser(); |
| 162 | } |
| 163 | |
| 164 | /** |
| 165 | * free the XML parser. |
| 166 | * |
| 167 | * @note It is unclear to me if we really need to do this ourselves |
| 168 | * or if php garbage collection will automatically free the xmlParser |
| 169 | * when it is no longer needed. |
| 170 | */ |
| 171 | private function destroyXMLParser(): void { |
| 172 | if ( $this->xmlParser ) { |
| 173 | xml_parser_free( $this->xmlParser ); |
| 174 | $this->xmlParser = null; |
| 175 | } |
| 176 | } |
| 177 | |
| 178 | /** |
| 179 | * Main use is if a single item has multiple xmp documents describing it. |
| 180 | * For example in jpeg's with extendedXMP |
| 181 | */ |
| 182 | private function resetXMLParser(): void { |
| 183 | $this->destroyXMLParser(); |
| 184 | |
| 185 | $this->xmlParser = xml_parser_create_ns( 'UTF-8', ' ' ); |
| 186 | xml_parser_set_option( $this->xmlParser, XML_OPTION_CASE_FOLDING, 0 ); |
| 187 | xml_parser_set_option( $this->xmlParser, XML_OPTION_SKIP_WHITE, 1 ); |
| 188 | |
| 189 | xml_set_element_handler( $this->xmlParser, |
| 190 | [ $this, 'startElement' ], |
| 191 | [ $this, 'endElement' ] ); |
| 192 | |
| 193 | xml_set_character_data_handler( $this->xmlParser, [ $this, 'char' ] ); |
| 194 | |
| 195 | $this->parsable = self::PARSABLE_UNKNOWN; |
| 196 | $this->xmlParsableBuffer = ''; |
| 197 | } |
| 198 | |
| 199 | /** |
| 200 | * Check if this instance supports using this class |
| 201 | * |
| 202 | * @return bool |
| 203 | */ |
| 204 | public static function isSupported(): bool { |
| 205 | return function_exists( 'xml_parser_create_ns' ) && class_exists( XMLReader::class ); |
| 206 | } |
| 207 | |
| 208 | /** |
| 209 | * Get the result array. Do some post-processing before returning |
| 210 | * the array, and transform any metadata that is special-cased. |
| 211 | * |
| 212 | * @return array Array of results as an array of arrays suitable for |
| 213 | * FormatMetadata::getFormattedData(). |
| 214 | */ |
| 215 | public function getResults(): array { |
| 216 | // xmp-special is for metadata that affects how stuff |
| 217 | // is extracted. For example xmpNote:HasExtendedXMP. |
| 218 | |
| 219 | // It is also used to handle photoshop:AuthorsPosition |
| 220 | // which is weird and really part of another property, |
| 221 | // see 2:85 in IPTC. See also pg 21 of IPTC4XMP standard. |
| 222 | // The location fields also use it. |
| 223 | |
| 224 | $data = $this->results; |
| 225 | |
| 226 | if ( isset( $data['xmp-special']['AuthorsPosition'] ) |
| 227 | && is_string( $data['xmp-special']['AuthorsPosition'] ) |
| 228 | && isset( $data['xmp-general']['Artist'][0] ) |
| 229 | ) { |
| 230 | // Note, if there is more than one creator, |
| 231 | // this only applies to first. This also will |
| 232 | // only apply to the dc:Creator prop, not the |
| 233 | // exif:Artist prop. |
| 234 | |
| 235 | $data['xmp-general']['Artist'][0] = |
| 236 | $data['xmp-special']['AuthorsPosition'] . ', ' |
| 237 | . $data['xmp-general']['Artist'][0]; |
| 238 | } |
| 239 | |
| 240 | // Go through the LocationShown and LocationCreated |
| 241 | // changing it to the non-hierarchical form used by |
| 242 | // the other location fields. |
| 243 | |
| 244 | if ( isset( $data['xmp-special']['LocationShown'][0] ) |
| 245 | // @phan-suppress-next-line PhanTypeArraySuspiciousNull, PhanTypeInvalidDimOffset |
| 246 | && is_array( $data['xmp-special']['LocationShown'][0] ) |
| 247 | ) { |
| 248 | // the is_array is just paranoia. It should always |
| 249 | // be an array. |
| 250 | foreach ( $data['xmp-special']['LocationShown'] as $loc ) { |
| 251 | if ( !is_array( $loc ) ) { |
| 252 | // To avoid copying over the _type meta-fields. |
| 253 | continue; |
| 254 | } |
| 255 | foreach ( $loc as $field => $val ) { |
| 256 | $data['xmp-general'][$field . 'Dest'][] = $val; |
| 257 | } |
| 258 | } |
| 259 | } |
| 260 | if ( isset( $data['xmp-special']['LocationCreated'][0] ) |
| 261 | // @phan-suppress-next-line PhanTypeArraySuspiciousNull, PhanTypeInvalidDimOffset |
| 262 | && is_array( $data['xmp-special']['LocationCreated'][0] ) |
| 263 | ) { |
| 264 | // the is_array is just paranoia. It should always |
| 265 | // be an array. |
| 266 | foreach ( $data['xmp-special']['LocationCreated'] as $loc ) { |
| 267 | if ( !is_array( $loc ) ) { |
| 268 | // To avoid copying over the _type meta-fields. |
| 269 | continue; |
| 270 | } |
| 271 | foreach ( $loc as $field => $val ) { |
| 272 | $data['xmp-general'][$field . 'Created'][] = $val; |
| 273 | } |
| 274 | } |
| 275 | } |
| 276 | |
| 277 | // We don't want to return the special values, since they're |
| 278 | // special and not info to be stored about the file. |
| 279 | unset( $data['xmp-special'] ); |
| 280 | |
| 281 | // Convert GPSAltitude to negative if below sea level. |
| 282 | if ( isset( $data['xmp-exif']['GPSAltitudeRef'] ) |
| 283 | && isset( $data['xmp-exif']['GPSAltitude'] ) |
| 284 | ) { |
| 285 | // Must convert to a real before multiplying by -1 |
| 286 | // Validate guarantees there will always be a '/' in this value. |
| 287 | [ $nom, $denom ] = explode( '/', $data['xmp-exif']['GPSAltitude'] ); |
| 288 | // @phan-suppress-next-line PhanTypeInvalidLeftOperandOfNumericOp, PhanTypeInvalidRightOperandOfNumericOp |
| 289 | $data['xmp-exif']['GPSAltitude'] = $nom / $denom; |
| 290 | |
| 291 | // @phan-suppress-next-line PhanTypeInvalidDimOffset |
| 292 | if ( $data['xmp-exif']['GPSAltitudeRef'] == '1' ) { |
| 293 | $data['xmp-exif']['GPSAltitude'] *= -1; |
| 294 | } |
| 295 | unset( $data['xmp-exif']['GPSAltitudeRef'] ); |
| 296 | } |
| 297 | |
| 298 | return $data; |
| 299 | } |
| 300 | |
| 301 | /** |
| 302 | * Main function to parse XMP. Use getResults to get results. |
| 303 | * |
| 304 | * Also catches any errors during processing, writes them to |
| 305 | * debug log, blanks result array and returns false. |
| 306 | * |
| 307 | * @param string $content XMP data |
| 308 | * @param bool $allOfIt If this is all the data (true), or if it's split up (false). Default true |
| 309 | * @throws RuntimeException |
| 310 | * @return bool Success. |
| 311 | */ |
| 312 | public function parse( $content, $allOfIt = true ): bool { |
| 313 | if ( !$this->xmlParser ) { |
| 314 | $this->resetXMLParser(); |
| 315 | } |
| 316 | try { |
| 317 | |
| 318 | // detect encoding by looking for BOM which is supposed to be in processing instruction. |
| 319 | // see page 12 of http://www.adobe.com/devnet/xmp/pdfs/XMPSpecificationPart3.pdf |
| 320 | if ( !$this->charset ) { |
| 321 | $bom = []; |
| 322 | if ( preg_match( '/\xEF\xBB\xBF|\xFE\xFF|\x00\x00\xFE\xFF|\xFF\xFE\x00\x00|\xFF\xFE/', |
| 323 | $content, $bom ) |
| 324 | ) { |
| 325 | switch ( $bom[0] ) { |
| 326 | case "\xFE\xFF": |
| 327 | $this->charset = 'UTF-16BE'; |
| 328 | break; |
| 329 | case "\xFF\xFE": |
| 330 | $this->charset = 'UTF-16LE'; |
| 331 | break; |
| 332 | case "\x00\x00\xFE\xFF": |
| 333 | $this->charset = 'UTF-32BE'; |
| 334 | break; |
| 335 | case "\xFF\xFE\x00\x00": |
| 336 | $this->charset = 'UTF-32LE'; |
| 337 | break; |
| 338 | case "\xEF\xBB\xBF": |
| 339 | $this->charset = 'UTF-8'; |
| 340 | break; |
| 341 | default: |
| 342 | // this should be impossible to get to |
| 343 | throw new RuntimeException( "Invalid BOM" ); |
| 344 | } |
| 345 | } else { |
| 346 | // standard specifically says, if no bom assume utf-8 |
| 347 | $this->charset = 'UTF-8'; |
| 348 | } |
| 349 | } |
| 350 | if ( $this->charset !== 'UTF-8' ) { |
| 351 | // don't convert if already utf-8 |
| 352 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 353 | $content = @iconv( $this->charset, 'UTF-8//IGNORE', $content ); |
| 354 | } |
| 355 | |
| 356 | // Replace any null bytes with the replacement character (T320282) |
| 357 | $content = str_replace( "\0", "\u{FFFD}", $content ); |
| 358 | |
| 359 | // Ensure the XMP block does not have an xml doctype declaration, which |
| 360 | // could declare entities unsafe to parse with xml_parse (T85848/T71210). |
| 361 | if ( $this->parsable !== self::PARSABLE_OK ) { |
| 362 | if ( $this->parsable === self::PARSABLE_NO ) { |
| 363 | throw new RuntimeException( 'Unsafe doctype declaration in XML.' ); |
| 364 | } |
| 365 | |
| 366 | $content = $this->xmlParsableBuffer . $content; |
| 367 | if ( !$this->checkParseSafety( $content ) ) { |
| 368 | if ( !$allOfIt && $this->parsable !== self::PARSABLE_NO ) { |
| 369 | // parse wasn't Unsuccessful yet, so return true |
| 370 | // in this case. |
| 371 | return true; |
| 372 | } |
| 373 | $msg = ( $this->parsable === self::PARSABLE_NO ) ? |
| 374 | 'Unsafe doctype declaration in XML.' : |
| 375 | 'No root element found in XML.'; |
| 376 | throw new RuntimeException( $msg ); |
| 377 | } |
| 378 | } |
| 379 | |
| 380 | $ok = xml_parse( $this->xmlParser, $content, $allOfIt ); |
| 381 | if ( !$ok ) { |
| 382 | $code = xml_get_error_code( $this->xmlParser ); |
| 383 | $error = xml_error_string( $code ); |
| 384 | $line = xml_get_current_line_number( $this->xmlParser ); |
| 385 | $col = xml_get_current_column_number( $this->xmlParser ); |
| 386 | $offset = xml_get_current_byte_index( $this->xmlParser ); |
| 387 | |
| 388 | $this->logger->info( |
| 389 | '{method} : Error reading XMP content: {error} ' . |
| 390 | '(file: {file}, line: {line} column: {column} ' . |
| 391 | 'byte offset: {offset})', |
| 392 | [ |
| 393 | 'method' => __METHOD__, |
| 394 | 'error_code' => $code, |
| 395 | 'error' => $error, |
| 396 | 'file' => $this->filename, |
| 397 | 'line' => $line, |
| 398 | 'column' => $col, |
| 399 | 'offset' => $offset, |
| 400 | 'content' => $content, |
| 401 | ] |
| 402 | ); |
| 403 | // blank if error. |
| 404 | $this->results = []; |
| 405 | $this->destroyXMLParser(); |
| 406 | return false; |
| 407 | } |
| 408 | } catch ( Exception $e ) { |
| 409 | $this->logger->warning( |
| 410 | '{method} {message}', |
| 411 | [ |
| 412 | 'method' => __METHOD__, |
| 413 | 'message' => $e->getMessage(), |
| 414 | 'exception' => $e, |
| 415 | 'file' => $this->filename, |
| 416 | 'content' => $content, |
| 417 | ] |
| 418 | ); |
| 419 | $this->results = []; |
| 420 | return false; |
| 421 | } |
| 422 | if ( $allOfIt ) { |
| 423 | $this->destroyXMLParser(); |
| 424 | } |
| 425 | |
| 426 | return true; |
| 427 | } |
| 428 | |
| 429 | /** Entry point for XMPExtended blocks in jpeg files |
| 430 | * |
| 431 | * @todo In serious need of testing |
| 432 | * @see http://www.adobe.ge/devnet/xmp/pdfs/XMPSpecificationPart3.pdf XMP spec part 3 page 20 |
| 433 | * @param string $content XMPExtended block minus the namespace signature |
| 434 | * @return bool If it succeeded. |
| 435 | */ |
| 436 | public function parseExtended( $content ): bool { |
| 437 | // @todo FIXME: This is untested. Hard to find example files |
| 438 | // or programs that make such files.. |
| 439 | $guid = substr( $content, 0, 32 ); |
| 440 | if ( !isset( $this->results['xmp-special']['HasExtendedXMP'] ) |
| 441 | || $this->results['xmp-special']['HasExtendedXMP'] !== $guid |
| 442 | ) { |
| 443 | $this->logger->info( __METHOD__ . |
| 444 | " Ignoring XMPExtended block due to wrong guid (guid= '{guid}')", |
| 445 | [ |
| 446 | 'guid' => $guid, |
| 447 | 'file' => $this->filename, |
| 448 | ] ); |
| 449 | |
| 450 | return false; |
| 451 | } |
| 452 | $len = unpack( 'Nlength/Noffset', substr( $content, 32, 8 ) ); |
| 453 | |
| 454 | if ( !$len || |
| 455 | $len['length'] < 4 || |
| 456 | $len['offset'] < 0 || |
| 457 | $len['offset'] > $len['length'] |
| 458 | ) { |
| 459 | $this->logger->info( |
| 460 | __METHOD__ . 'Error reading extended XMP block, invalid length or offset.', |
| 461 | [ 'file' => $this->filename ] |
| 462 | ); |
| 463 | |
| 464 | return false; |
| 465 | } |
| 466 | |
| 467 | // we're not very robust here. we should accept it in the wrong order. |
| 468 | // To quote the XMP standard: |
| 469 | // "A JPEG writer should write the ExtendedXMP marker segments in order, |
| 470 | // immediately following the StandardXMP. However, the JPEG standard |
| 471 | // does not require preservation of marker segment order. A robust JPEG |
| 472 | // reader should tolerate the marker segments in any order." |
| 473 | // On the other hand, the probability that an image will have more than |
| 474 | // 128k of metadata is rather low... so the probability that it will have |
| 475 | // > 128k, and be in the wrong order is very low... |
| 476 | |
| 477 | if ( $len['offset'] !== $this->extendedXMPOffset ) { |
| 478 | $this->logger->info( __METHOD__ . 'Ignoring XMPExtended block due to wrong order. (Offset was ' |
| 479 | . $len['offset'] . ' but expected ' . $this->extendedXMPOffset . ')', |
| 480 | [ 'file' => $this->filename ] |
| 481 | ); |
| 482 | |
| 483 | return false; |
| 484 | } |
| 485 | |
| 486 | if ( $len['offset'] === 0 ) { |
| 487 | // if we're starting the extended block, we've probably already |
| 488 | // done the XMPStandard block, so reset. |
| 489 | $this->resetXMLParser(); |
| 490 | } |
| 491 | |
| 492 | $this->extendedXMPOffset += $len['length']; |
| 493 | |
| 494 | $actualContent = substr( $content, 40 ); |
| 495 | |
| 496 | $atEnd = ( $this->extendedXMPOffset === strlen( $actualContent ) ); |
| 497 | |
| 498 | $this->logger->debug( |
| 499 | __METHOD__ . 'Parsing a XMPExtended block', |
| 500 | [ 'file' => $this->filename ] |
| 501 | ); |
| 502 | |
| 503 | return $this->parse( $actualContent, $atEnd ); |
| 504 | } |
| 505 | |
| 506 | /** |
| 507 | * Character data handler |
| 508 | * Called whenever character data is found in the xmp document. |
| 509 | * |
| 510 | * does nothing if we're in MODE_IGNORE or if the data is whitespace |
| 511 | * throws an error if we're not in MODE_SIMPLE (as we're not allowed to have character |
| 512 | * data in the other modes). |
| 513 | * |
| 514 | * As an example, this happens when we encounter XMP like: |
| 515 | * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio> |
| 516 | * and are processing the 0/10 bit. |
| 517 | * |
| 518 | * @param resource $parser XMLParser reference to the xml parser |
| 519 | * @param string $data Character data |
| 520 | * @throws RuntimeException On invalid data |
| 521 | */ |
| 522 | public function char( $parser, $data ): void { |
| 523 | $data = trim( $data ); |
| 524 | if ( trim( $data ) === "" ) { |
| 525 | return; |
| 526 | } |
| 527 | |
| 528 | if ( !isset( $this->mode[0] ) ) { |
| 529 | throw new RuntimeException( 'Unexpected character data before first rdf:Description element' ); |
| 530 | } |
| 531 | |
| 532 | if ( $this->mode[0] === self::MODE_IGNORE ) { |
| 533 | return; |
| 534 | } |
| 535 | |
| 536 | if ( $this->mode[0] !== self::MODE_SIMPLE |
| 537 | && $this->mode[0] !== self::MODE_QDESC |
| 538 | ) { |
| 539 | throw new RuntimeException( 'character data where not expected. (mode ' . $this->mode[0] . ')' ); |
| 540 | } |
| 541 | |
| 542 | // to check, how does this handle w.s. |
| 543 | if ( $this->charContent === false ) { |
| 544 | $this->charContent = $data; |
| 545 | } else { |
| 546 | $this->charContent .= $data; |
| 547 | } |
| 548 | } |
| 549 | |
| 550 | /** |
| 551 | * Check if a block of XML is safe to pass to xml_parse, i.e. doesn't |
| 552 | * contain a doctype declaration which could contain a dos attack if we |
| 553 | * parse it and expand internal entities (T85848). |
| 554 | * |
| 555 | * @param string $content xml string to check for parse safety |
| 556 | * @return bool true if the xml is safe to parse, false otherwise |
| 557 | */ |
| 558 | private function checkParseSafety( $content ): bool { |
| 559 | $reader = new XMLReader(); |
| 560 | $result = null; |
| 561 | |
| 562 | // Pull in the arbitrary MAX_URI_LENGTH from libxml2... |
| 563 | $maxUriLength = 1024 * 1024; |
| 564 | $dataUri = 'data://text/plain,' . urlencode( $content ); |
| 565 | if ( strlen( $dataUri ) > $maxUriLength ) { |
| 566 | // libxml2 won't parse this file as a data URI due to the length. |
| 567 | return false; |
| 568 | } |
| 569 | |
| 570 | // For XMLReader to parse incomplete/invalid XML, it has to be open()'ed |
| 571 | // instead of using XML(). |
| 572 | if ( !$reader->open( |
| 573 | $dataUri, |
| 574 | null, |
| 575 | LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NONET |
| 576 | ) ) { |
| 577 | return false; |
| 578 | } |
| 579 | |
| 580 | if ( LIBXML_VERSION < 20900 ) { |
| 581 | $oldDisable = libxml_disable_entity_loader( true ); |
| 582 | /** @noinspection PhpUnusedLocalVariableInspection */ |
| 583 | $reset = new ScopedCallback( |
| 584 | 'libxml_disable_entity_loader', |
| 585 | [ $oldDisable ] |
| 586 | ); |
| 587 | } |
| 588 | |
| 589 | $reader->setParserProperty( XMLReader::SUBST_ENTITIES, false ); |
| 590 | |
| 591 | // Even with LIBXML_NOWARNING set, XMLReader::read gives a warning |
| 592 | // when parsing truncated XML, which causes unit tests to fail. |
| 593 | // phpcs:ignore Generic.PHP.NoSilencedErrors.Discouraged |
| 594 | while ( @$reader->read() ) { |
| 595 | if ( $reader->nodeType === XMLReader::ELEMENT ) { |
| 596 | // Reached the first element without hitting a doctype declaration |
| 597 | $this->parsable = self::PARSABLE_OK; |
| 598 | $result = true; |
| 599 | break; |
| 600 | } |
| 601 | if ( $reader->nodeType === XMLReader::DOC_TYPE ) { |
| 602 | $this->parsable = self::PARSABLE_NO; |
| 603 | $result = false; |
| 604 | break; |
| 605 | } |
| 606 | } |
| 607 | |
| 608 | if ( $result !== null ) { |
| 609 | return $result; |
| 610 | } |
| 611 | |
| 612 | // Reached the end of the parsable xml without finding an element |
| 613 | // or doctype. Buffer and try again. |
| 614 | $this->parsable = self::PARSABLE_BUFFERING; |
| 615 | $this->xmlParsableBuffer = $content; |
| 616 | return false; |
| 617 | } |
| 618 | |
| 619 | /** When we hit a closing element in MODE_IGNORE |
| 620 | * Check to see if this is the element we started to ignore, |
| 621 | * in which case we get out of MODE_IGNORE |
| 622 | * |
| 623 | * @param string $elm Namespace of element followed by a space and then tag name of element. |
| 624 | */ |
| 625 | private function endElementModeIgnore( $elm ): void { |
| 626 | if ( $this->curItem[0] === $elm ) { |
| 627 | array_shift( $this->curItem ); |
| 628 | array_shift( $this->mode ); |
| 629 | } |
| 630 | } |
| 631 | |
| 632 | /** |
| 633 | * Hit a closing element when in MODE_SIMPLE. |
| 634 | * This generally means that we finished processing a |
| 635 | * property value, and now have to save the result to the |
| 636 | * results array |
| 637 | * |
| 638 | * For example, when processing: |
| 639 | * <exif:DigitalZoomRatio>0/10</exif:DigitalZoomRatio> |
| 640 | * this deals with when we hit </exif:DigitalZoomRatio>. |
| 641 | * |
| 642 | * Or it could be if we hit the end element of a property |
| 643 | * of a compound data structure (like a member of an array). |
| 644 | * |
| 645 | * @param string $elm Namespace, space, and tag name. |
| 646 | */ |
| 647 | private function endElementModeSimple( $elm ): void { |
| 648 | if ( $this->charContent !== false ) { |
| 649 | if ( $this->processingArray ) { |
| 650 | // if we're processing an array, use the original element |
| 651 | // name instead of rdf:li. |
| 652 | [ $ns, $tag ] = explode( ' ', $this->curItem[0], 2 ); |
| 653 | } else { |
| 654 | [ $ns, $tag ] = explode( ' ', $elm, 2 ); |
| 655 | } |
| 656 | $this->saveValue( $ns, $tag, $this->charContent ); |
| 657 | |
| 658 | // reset |
| 659 | $this->charContent = false; |
| 660 | } |
| 661 | array_shift( $this->curItem ); |
| 662 | array_shift( $this->mode ); |
| 663 | } |
| 664 | |
| 665 | /** |
| 666 | * Hit a closing element in MODE_STRUCT, MODE_SEQ, MODE_BAG |
| 667 | * generally means we've finished processing a nested structure. |
| 668 | * resets some internal variables to indicate that. |
| 669 | * |
| 670 | * Note this means we hit the closing element not the "</rdf:Seq>". |
| 671 | * |
| 672 | * @par For example, when processing: |
| 673 | * @code{.xml} |
| 674 | * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li> |
| 675 | * </rdf:Seq> </exif:ISOSpeedRatings> |
| 676 | * @endcode |
| 677 | * |
| 678 | * This method is called when we hit the "</exif:ISOSpeedRatings>" tag. |
| 679 | * |
| 680 | * @param string $elm Namespace . space . tag name. |
| 681 | * @throws RuntimeException |
| 682 | */ |
| 683 | private function endElementNested( $elm ): void { |
| 684 | /* cur item must be the same as $elm, unless if in MODE_STRUCT |
| 685 | * in which case it could also be rdf:Description */ |
| 686 | if ( $this->curItem[0] !== $elm |
| 687 | && !( $elm === self::NS_RDF . ' Description' |
| 688 | && $this->mode[0] === self::MODE_STRUCT ) |
| 689 | ) { |
| 690 | throw new RuntimeException( "nesting mismatch. got a </$elm> but expected a </" . |
| 691 | $this->curItem[0] . '>' ); |
| 692 | } |
| 693 | |
| 694 | // Validate structures. |
| 695 | [ $ns, $tag ] = explode( ' ', $elm, 2 ); |
| 696 | if ( isset( $this->items[$ns][$tag]['validate'] ) ) { |
| 697 | $info =& $this->items[$ns][$tag]; |
| 698 | $finalName = $info['map_name'] ?? $tag; |
| 699 | |
| 700 | if ( is_array( $info['validate'] ) ) { |
| 701 | $validate = $info['validate']; |
| 702 | } else { |
| 703 | $validator = new Validate( $this->logger ); |
| 704 | $validate = [ $validator, $info['validate'] ]; |
| 705 | } |
| 706 | |
| 707 | if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) { |
| 708 | // This can happen if all the members of the struct failed validation. |
| 709 | $this->logger->debug( |
| 710 | __METHOD__ . " <$ns:$tag> has no valid members.", |
| 711 | [ 'file' => $this->filename ] |
| 712 | ); |
| 713 | } elseif ( is_callable( $validate ) ) { |
| 714 | $val =& $this->results['xmp-' . $info['map_group']][$finalName]; |
| 715 | call_user_func_array( $validate, [ $info, &$val, false ] ); |
| 716 | if ( $val === null ) { |
| 717 | // the idea being the validation function will unset the variable if |
| 718 | // its invalid. |
| 719 | $this->logger->info( |
| 720 | __METHOD__ . " <$ns:$tag> failed validation.", |
| 721 | [ 'file' => $this->filename ] |
| 722 | ); |
| 723 | unset( $this->results['xmp-' . $info['map_group']][$finalName] ); |
| 724 | } |
| 725 | } else { |
| 726 | $this->logger->warning( |
| 727 | __METHOD__ . " Validation function for $finalName (" . |
| 728 | get_class( $validate[0] ) . '::' . $validate[1] . '()) is not callable.', |
| 729 | [ 'file' => $this->filename ] |
| 730 | ); |
| 731 | } |
| 732 | } |
| 733 | |
| 734 | array_shift( $this->curItem ); |
| 735 | array_shift( $this->mode ); |
| 736 | $this->ancestorStruct = false; |
| 737 | $this->processingArray = false; |
| 738 | $this->itemLang = false; |
| 739 | } |
| 740 | |
| 741 | /** |
| 742 | * Hit a closing element in MODE_LI (either rdf:Seq, or rdf:Bag ) |
| 743 | * Add information about what type of element this is. |
| 744 | * |
| 745 | * Note we still have to hit the outer "</property>" |
| 746 | * |
| 747 | * @par For example, when processing: |
| 748 | * @code{.xml} |
| 749 | * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li> |
| 750 | * </rdf:Seq> </exif:ISOSpeedRatings> |
| 751 | * @endcode |
| 752 | * |
| 753 | * This method is called when we hit the "</rdf:Seq>". |
| 754 | * (For comparison, we call endElementModeSimple when we |
| 755 | * hit the "</rdf:li>") |
| 756 | * |
| 757 | * @param string $elm Namespace . ' ' . element name |
| 758 | * @throws RuntimeException |
| 759 | */ |
| 760 | private function endElementModeLi( $elm ): void { |
| 761 | [ $ns, $tag ] = explode( ' ', $this->curItem[0], 2 ); |
| 762 | $info = $this->items[$ns][$tag]; |
| 763 | $finalName = $info['map_name'] ?? $tag; |
| 764 | |
| 765 | array_shift( $this->mode ); |
| 766 | |
| 767 | if ( !isset( $this->results['xmp-' . $info['map_group']][$finalName] ) ) { |
| 768 | $this->logger->debug( |
| 769 | __METHOD__ . " Empty compound element $finalName.", |
| 770 | [ 'file' => $this->filename ] |
| 771 | ); |
| 772 | |
| 773 | return; |
| 774 | } |
| 775 | |
| 776 | if ( $elm === self::NS_RDF . ' Seq' ) { |
| 777 | $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ol'; |
| 778 | } elseif ( $elm === self::NS_RDF . ' Bag' ) { |
| 779 | $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'ul'; |
| 780 | } elseif ( $elm === self::NS_RDF . ' Alt' ) { |
| 781 | // extra if needed as you could theoretically have a non-language alt. |
| 782 | if ( $info['mode'] === self::MODE_LANG ) { |
| 783 | $this->results['xmp-' . $info['map_group']][$finalName]['_type'] = 'lang'; |
| 784 | } |
| 785 | } else { |
| 786 | throw new RuntimeException( |
| 787 | __METHOD__ . " expected </rdf:seq> or </rdf:bag> but instead got $elm." |
| 788 | ); |
| 789 | } |
| 790 | } |
| 791 | |
| 792 | /** |
| 793 | * End element while in MODE_QDESC |
| 794 | * mostly when ending an element when we have a simple value |
| 795 | * that has qualifiers. |
| 796 | * |
| 797 | * Qualifiers aren't all that common, and we don't do anything |
| 798 | * with them. |
| 799 | * |
| 800 | * @param string $elm Namespace and element |
| 801 | */ |
| 802 | private function endElementModeQDesc( $elm ): void { |
| 803 | if ( $elm === self::NS_RDF . ' value' ) { |
| 804 | [ $ns, $tag ] = explode( ' ', $this->curItem[0], 2 ); |
| 805 | $this->saveValue( $ns, $tag, $this->charContent ); |
| 806 | |
| 807 | return; |
| 808 | } |
| 809 | |
| 810 | array_shift( $this->mode ); |
| 811 | array_shift( $this->curItem ); |
| 812 | } |
| 813 | |
| 814 | /** |
| 815 | * Handler for hitting a closing element. |
| 816 | * |
| 817 | * generally just calls a helper function depending on what |
| 818 | * mode we're in. |
| 819 | * |
| 820 | * Ignores the outer wrapping elements that are optional in |
| 821 | * xmp and have no meaning. |
| 822 | * |
| 823 | * @param resource $parser |
| 824 | * @param string $elm Namespace . ' ' . element name |
| 825 | * @throws RuntimeException |
| 826 | */ |
| 827 | public function endElement( $parser, $elm ): void { |
| 828 | if ( $elm === ( self::NS_RDF . ' RDF' ) |
| 829 | || $elm === 'adobe:ns:meta/ xmpmeta' |
| 830 | || $elm === 'adobe:ns:meta/ xapmeta' |
| 831 | ) { |
| 832 | // ignore these. |
| 833 | return; |
| 834 | } |
| 835 | |
| 836 | if ( $elm === self::NS_RDF . ' type' ) { |
| 837 | // these aren't really supported properly yet. |
| 838 | // However, it appears they almost never used. |
| 839 | $this->logger->info( |
| 840 | __METHOD__ . ' encountered <rdf:type>', |
| 841 | [ 'file' => $this->filename ] |
| 842 | ); |
| 843 | } |
| 844 | |
| 845 | if ( strpos( $elm, ' ' ) === false ) { |
| 846 | // This probably shouldn't happen. |
| 847 | // However, there is a bug in an adobe product |
| 848 | // that forgets the namespace on some things. |
| 849 | // (Luckily they are unimportant things). |
| 850 | $this->logger->info( |
| 851 | __METHOD__ . " Encountered </$elm> which has no namespace. Skipping.", |
| 852 | [ 'file' => $this->filename ] |
| 853 | ); |
| 854 | |
| 855 | return; |
| 856 | } |
| 857 | |
| 858 | if ( count( $this->mode ) === 0 ) { |
| 859 | // This should never ever happen and means |
| 860 | // there is a pretty major bug in this class. |
| 861 | throw new RuntimeException( 'Encountered end element with no mode' ); |
| 862 | } |
| 863 | |
| 864 | if ( count( $this->curItem ) === 0 && $this->mode[0] !== self::MODE_INITIAL ) { |
| 865 | // just to be paranoid. Should always have a curItem, except for initially |
| 866 | // (aka during MODE_INITIAL). |
| 867 | throw new RuntimeException( "Hit end element </$elm> but no curItem" ); |
| 868 | } |
| 869 | |
| 870 | switch ( $this->mode[0] ) { |
| 871 | case self::MODE_IGNORE: |
| 872 | $this->endElementModeIgnore( $elm ); |
| 873 | break; |
| 874 | case self::MODE_SIMPLE: |
| 875 | $this->endElementModeSimple( $elm ); |
| 876 | break; |
| 877 | case self::MODE_STRUCT: |
| 878 | case self::MODE_SEQ: |
| 879 | case self::MODE_BAG: |
| 880 | case self::MODE_LANG: |
| 881 | case self::MODE_BAGSTRUCT: |
| 882 | $this->endElementNested( $elm ); |
| 883 | break; |
| 884 | case self::MODE_INITIAL: |
| 885 | if ( $elm === self::NS_RDF . ' Description' ) { |
| 886 | array_shift( $this->mode ); |
| 887 | } else { |
| 888 | throw new RuntimeException( 'Element ended unexpectedly while in MODE_INITIAL' ); |
| 889 | } |
| 890 | break; |
| 891 | case self::MODE_LI: |
| 892 | case self::MODE_LI_LANG: |
| 893 | $this->endElementModeLi( $elm ); |
| 894 | break; |
| 895 | case self::MODE_QDESC: |
| 896 | $this->endElementModeQDesc( $elm ); |
| 897 | break; |
| 898 | default: |
| 899 | $this->logger->info( |
| 900 | __METHOD__ . " no mode (elm = $elm)", |
| 901 | [ 'file' => $this->filename ] |
| 902 | ); |
| 903 | break; |
| 904 | } |
| 905 | } |
| 906 | |
| 907 | /** |
| 908 | * Hit an opening element while in MODE_IGNORE |
| 909 | * |
| 910 | * XMP is extensible, so ignore any tag we don't understand. |
| 911 | * |
| 912 | * Mostly ignores, unless we encounter the element that we are ignoring. |
| 913 | * in which case we add it to the item stack, so we can ignore things |
| 914 | * that are nested, correctly. |
| 915 | * |
| 916 | * @param string $elm Namespace . ' ' . tag name |
| 917 | */ |
| 918 | private function startElementModeIgnore( $elm ): void { |
| 919 | if ( $elm === $this->curItem[0] ) { |
| 920 | array_unshift( $this->curItem, $elm ); |
| 921 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 922 | } |
| 923 | } |
| 924 | |
| 925 | /** |
| 926 | * Start element in MODE_BAG (unordered array) |
| 927 | * this should always be <rdf:Bag> |
| 928 | * |
| 929 | * @param string $elm Namespace . ' ' . tag |
| 930 | * @throws RuntimeException If we have an element that's not <rdf:Bag> |
| 931 | */ |
| 932 | private function startElementModeBag( $elm ): void { |
| 933 | if ( $elm === self::NS_RDF . ' Bag' ) { |
| 934 | array_unshift( $this->mode, self::MODE_LI ); |
| 935 | } else { |
| 936 | throw new RuntimeException( "Expected <rdf:Bag> but got $elm." ); |
| 937 | } |
| 938 | } |
| 939 | |
| 940 | /** |
| 941 | * Start element in MODE_SEQ (ordered array) |
| 942 | * this should always be <rdf:Seq> |
| 943 | * |
| 944 | * @param string $elm Namespace . ' ' . tag |
| 945 | * @throws RuntimeException If we have an element that's not <rdf:Seq> |
| 946 | */ |
| 947 | private function startElementModeSeq( $elm ): void { |
| 948 | if ( $elm === self::NS_RDF . ' Seq' ) { |
| 949 | array_unshift( $this->mode, self::MODE_LI ); |
| 950 | } elseif ( $elm === self::NS_RDF . ' Bag' ) { |
| 951 | # T29105 |
| 952 | $this->logger->info( |
| 953 | __METHOD__ . ' Expected an rdf:Seq, but got an rdf:Bag. Pretending' . |
| 954 | ' it is a Seq, since some buggy software is known to screw this up.', |
| 955 | [ 'file' => $this->filename ] |
| 956 | ); |
| 957 | array_unshift( $this->mode, self::MODE_LI ); |
| 958 | } else { |
| 959 | throw new RuntimeException( "Expected <rdf:Seq> but got $elm." ); |
| 960 | } |
| 961 | } |
| 962 | |
| 963 | /** |
| 964 | * Start element in MODE_LANG (language alternative) |
| 965 | * this should always be <rdf:Alt> |
| 966 | * |
| 967 | * This tag tends to be used for metadata like describe this |
| 968 | * picture, which can be translated into multiple languages. |
| 969 | * |
| 970 | * XMP supports non-linguistic alternative selections, |
| 971 | * which are really only used for thumbnails, which |
| 972 | * we don't care about. |
| 973 | * |
| 974 | * @param string $elm Namespace . ' ' . tag |
| 975 | * @throws RuntimeException If we have an element that's not <rdf:Alt> |
| 976 | */ |
| 977 | private function startElementModeLang( $elm ): void { |
| 978 | if ( $elm === self::NS_RDF . ' Alt' ) { |
| 979 | array_unshift( $this->mode, self::MODE_LI_LANG ); |
| 980 | } else { |
| 981 | throw new RuntimeException( "Expected <rdf:Seq> but got $elm." ); |
| 982 | } |
| 983 | } |
| 984 | |
| 985 | /** |
| 986 | * Handle an opening element when in MODE_SIMPLE |
| 987 | * |
| 988 | * This should not happen often. This is for if a simple element |
| 989 | * already opened has a child element. Could happen for a |
| 990 | * qualified element. |
| 991 | * |
| 992 | * For example: |
| 993 | * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value> |
| 994 | * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description> |
| 995 | * </exif:DigitalZoomRatio> |
| 996 | * |
| 997 | * This method is called when processing the <rdf:Description> element |
| 998 | * |
| 999 | * @param string $elm Namespace and tag names separated by space. |
| 1000 | * @param array $attribs Attributes of the element. |
| 1001 | * @throws RuntimeException |
| 1002 | */ |
| 1003 | private function startElementModeSimple( $elm, $attribs ): void { |
| 1004 | if ( $elm === self::NS_RDF . ' Description' ) { |
| 1005 | // If this value has qualifiers |
| 1006 | array_unshift( $this->mode, self::MODE_QDESC ); |
| 1007 | array_unshift( $this->curItem, $this->curItem[0] ); |
| 1008 | |
| 1009 | if ( isset( $attribs[self::NS_RDF . ' value'] ) ) { |
| 1010 | [ $ns, $tag ] = explode( ' ', $this->curItem[0], 2 ); |
| 1011 | $this->saveValue( $ns, $tag, $attribs[self::NS_RDF . ' value'] ); |
| 1012 | } |
| 1013 | } elseif ( $elm === self::NS_RDF . ' value' ) { |
| 1014 | // This should not be here. |
| 1015 | throw new RuntimeException( __METHOD__ . ' Encountered <rdf:value> where it was unexpected.' ); |
| 1016 | } else { |
| 1017 | // something else we don't recognize, like a qualifier maybe. |
| 1018 | $this->logger->info( __METHOD__ . |
| 1019 | " Encountered element <{element}> where only expecting character data as value of {curitem}", |
| 1020 | [ |
| 1021 | 'element' => $elm, |
| 1022 | 'curitem' => $this->curItem[0], |
| 1023 | 'file' => $this->filename, |
| 1024 | ] |
| 1025 | ); |
| 1026 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 1027 | array_unshift( $this->curItem, $elm ); |
| 1028 | } |
| 1029 | } |
| 1030 | |
| 1031 | /** |
| 1032 | * Start an element when in MODE_QDESC. |
| 1033 | * This generally happens when a simple element has an inner |
| 1034 | * rdf:Description to hold qualifier elements. |
| 1035 | * |
| 1036 | * For example in: |
| 1037 | * <exif:DigitalZoomRatio><rdf:Description><rdf:value>0/10</rdf:value> |
| 1038 | * <foo:someQualifier>Bar</foo:someQualifier> </rdf:Description> |
| 1039 | * </exif:DigitalZoomRatio> |
| 1040 | * Called when processing the <rdf:value> or <foo:someQualifier>. |
| 1041 | * |
| 1042 | * @param string $elm Namespace and tag name separated by a space. |
| 1043 | */ |
| 1044 | private function startElementModeQDesc( $elm ): void { |
| 1045 | if ( $elm === self::NS_RDF . ' value' ) { |
| 1046 | // do nothing |
| 1047 | return; |
| 1048 | } |
| 1049 | |
| 1050 | // otherwise its a qualifier, which we ignore |
| 1051 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 1052 | array_unshift( $this->curItem, $elm ); |
| 1053 | } |
| 1054 | |
| 1055 | /** |
| 1056 | * Starting an element when in MODE_INITIAL |
| 1057 | * This usually happens when we hit an element inside |
| 1058 | * the outer rdf:Description |
| 1059 | * |
| 1060 | * This is generally where most properties start. |
| 1061 | * |
| 1062 | * @param string $ns Namespace |
| 1063 | * @param string $tag Tag name (without namespace prefix) |
| 1064 | * @param array $attribs Array of attributes |
| 1065 | * @throws RuntimeException |
| 1066 | */ |
| 1067 | private function startElementModeInitial( $ns, $tag, $attribs ): void { |
| 1068 | if ( $ns !== self::NS_RDF ) { |
| 1069 | if ( isset( $this->items[$ns][$tag] ) ) { |
| 1070 | if ( isset( $this->items[$ns][$tag]['structPart'] ) ) { |
| 1071 | // If this element is supposed to appear only as |
| 1072 | // a child of a structure, but appears here (not as |
| 1073 | // a child of a struct), then something weird is |
| 1074 | // happening, so ignore this element and its children. |
| 1075 | |
| 1076 | $this->logger->info( |
| 1077 | 'Encountered <{element}> outside of its expected parent. Ignoring.', |
| 1078 | [ 'element' => "$ns:$tag", 'file' => $this->filename ] |
| 1079 | ); |
| 1080 | |
| 1081 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 1082 | array_unshift( $this->curItem, $ns . ' ' . $tag ); |
| 1083 | |
| 1084 | return; |
| 1085 | } |
| 1086 | $mode = $this->items[$ns][$tag]['mode']; |
| 1087 | array_unshift( $this->mode, $mode ); |
| 1088 | array_unshift( $this->curItem, $ns . ' ' . $tag ); |
| 1089 | if ( $mode === self::MODE_STRUCT ) { |
| 1090 | $this->ancestorStruct = $this->items[$ns][$tag]['map_name'] ?? $tag; |
| 1091 | } |
| 1092 | if ( $this->charContent !== false ) { |
| 1093 | // Something weird. |
| 1094 | // Should not happen in valid XMP. |
| 1095 | throw new RuntimeException( 'tag nested in non-whitespace characters.' ); |
| 1096 | } |
| 1097 | } else { |
| 1098 | // This element is not on our list of allowed elements so ignore. |
| 1099 | $this->logger->debug( __METHOD__ . ' Ignoring unrecognized element <{element}>.', |
| 1100 | [ 'element' => "$ns:$tag", 'file' => $this->filename ] ); |
| 1101 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 1102 | array_unshift( $this->curItem, $ns . ' ' . $tag ); |
| 1103 | |
| 1104 | return; |
| 1105 | } |
| 1106 | } |
| 1107 | // process attributes |
| 1108 | $this->doAttribs( $attribs ); |
| 1109 | } |
| 1110 | |
| 1111 | /** |
| 1112 | * Hit an opening element when in a Struct (MODE_STRUCT) |
| 1113 | * This is generally for fields of a compound property. |
| 1114 | * |
| 1115 | * Example of a struct (abbreviated; flash has more properties): |
| 1116 | * |
| 1117 | * <exif:Flash> <rdf:Description> <exif:Fired>True</exif:Fired> |
| 1118 | * <exif:Mode>1</exif:Mode></rdf:Description></exif:Flash> |
| 1119 | * |
| 1120 | * or: |
| 1121 | * |
| 1122 | * <exif:Flash rdf:parseType='Resource'> <exif:Fired>True</exif:Fired> |
| 1123 | * <exif:Mode>1</exif:Mode></exif:Flash> |
| 1124 | * |
| 1125 | * @param string $ns Namespace |
| 1126 | * @param string $tag Tag name (no ns) |
| 1127 | * @param array $attribs Array of attribs w/ values. |
| 1128 | * @throws RuntimeException |
| 1129 | */ |
| 1130 | private function startElementModeStruct( $ns, $tag, $attribs ): void { |
| 1131 | if ( $ns !== self::NS_RDF ) { |
| 1132 | if ( isset( $this->items[$ns][$tag] ) ) { |
| 1133 | if ( isset( $this->items[$ns][$this->ancestorStruct]['children'] ) |
| 1134 | && !isset( $this->items[$ns][$this->ancestorStruct]['children'][$tag] ) |
| 1135 | ) { |
| 1136 | // This assumes that we don't have inter-namespace nesting |
| 1137 | // which we don't in all the properties we're interested in. |
| 1138 | throw new RuntimeException( " <$tag> appeared nested in <" . $this->ancestorStruct |
| 1139 | . "> where it is not allowed." ); |
| 1140 | } |
| 1141 | array_unshift( $this->mode, $this->items[$ns][$tag]['mode'] ); |
| 1142 | array_unshift( $this->curItem, $ns . ' ' . $tag ); |
| 1143 | if ( $this->charContent !== false ) { |
| 1144 | // Something weird. |
| 1145 | // Should not happen in valid XMP. |
| 1146 | throw new RuntimeException( "tag <$tag> nested in non-whitespace characters (" . |
| 1147 | $this->charContent . ")." ); |
| 1148 | } |
| 1149 | } else { |
| 1150 | array_unshift( $this->mode, self::MODE_IGNORE ); |
| 1151 | array_unshift( $this->curItem, $ns . ' ' . $tag ); |
| 1152 | |
| 1153 | return; |
| 1154 | } |
| 1155 | } |
| 1156 | |
| 1157 | if ( $ns === self::NS_RDF && $tag === 'Description' ) { |
| 1158 | $this->doAttribs( $attribs ); |
| 1159 | array_unshift( $this->mode, self::MODE_STRUCT ); |
| 1160 | array_unshift( $this->curItem, $this->curItem[0] ); |
| 1161 | } |
| 1162 | } |
| 1163 | |
| 1164 | /** |
| 1165 | * opening element in MODE_LI |
| 1166 | * process elements of arrays. |
| 1167 | * |
| 1168 | * Example: |
| 1169 | * <exif:ISOSpeedRatings> <rdf:Seq> <rdf:li>64</rdf:li> |
| 1170 | * </rdf:Seq> </exif:ISOSpeedRatings> |
| 1171 | * This method is called when we hit the <rdf:li> element. |
| 1172 | * |
| 1173 | * @param string $elm Namespace . ' ' . tagname |
| 1174 | * @param array $attribs Attributes. (needed for BAGSTRUCTS) |
| 1175 | * @throws RuntimeException If it gets a tag other than <rdf:li> |
| 1176 | */ |
| 1177 | private function startElementModeLi( $elm, $attribs ): void { |
| 1178 | if ( ( $elm ) !== self::NS_RDF . ' li' ) { |
| 1179 | throw new RuntimeException( "<rdf:li> expected but got $elm." ); |
| 1180 | } |
| 1181 | |
| 1182 | if ( !isset( $this->mode[1] ) ) { |
| 1183 | // This should never ever ever happen. Checking for it |
| 1184 | // to be paranoid. |
| 1185 | throw new RuntimeException( 'In mode Li, but no 2xPrevious mode!' ); |
| 1186 | } |
| 1187 | |
| 1188 | if ( $this->mode[1] === self::MODE_BAGSTRUCT ) { |
| 1189 | // This list item contains a compound (STRUCT) value. |
| 1190 | array_unshift( $this->mode, self::MODE_STRUCT ); |
| 1191 | array_unshift( $this->curItem, $elm ); |
| 1192 | $this->processingArray = true; |
| 1193 | |
| 1194 | if ( !isset( $this->curItem[1] ) ) { |
| 1195 | // be paranoid. |
| 1196 | throw new RuntimeException( 'Can not find parent of BAGSTRUCT.' ); |
| 1197 | } |
| 1198 | [ $curNS, $curTag ] = explode( ' ', $this->curItem[1] ); |
| 1199 | $this->ancestorStruct = $this->items[$curNS][$curTag]['map_name'] ?? $curTag; |
| 1200 | |
| 1201 | $this->doAttribs( $attribs ); |
| 1202 | } else { |
| 1203 | // Normal BAG or SEQ containing simple values. |
| 1204 | array_unshift( $this->mode, self::MODE_SIMPLE ); |
| 1205 | // need to add curItem[0] on again since one is for the specific item |
| 1206 | // and one is for the entire group. |
| 1207 | array_unshift( $this->curItem, $this->curItem[0] ); |
| 1208 | $this->processingArray = true; |
| 1209 | } |
| 1210 | } |
| 1211 | |
| 1212 | /** |
| 1213 | * Opening element in MODE_LI_LANG. |
| 1214 | * process elements of language alternatives |
| 1215 | * |
| 1216 | * Example: |
| 1217 | * <dc:title> <rdf:Alt> <rdf:li xml:lang="x-default">My house |
| 1218 | * </rdf:li> </rdf:Alt> </dc:title> |
| 1219 | * |
| 1220 | * This method is called when we hit the <rdf:li> element. |
| 1221 | * |
| 1222 | * @param string $elm Namespace . ' ' . tag |
| 1223 | * @param array $attribs Array of elements (most importantly xml:lang) |
| 1224 | * @throws RuntimeException If it gets a tag other than <rdf:li> or if no xml:lang |
| 1225 | */ |
| 1226 | private function startElementModeLiLang( $elm, $attribs ): void { |
| 1227 | if ( $elm !== self::NS_RDF . ' li' ) { |
| 1228 | throw new RuntimeException( __METHOD__ . " <rdf:li> expected but got $elm." ); |
| 1229 | } |
| 1230 | if ( !isset( $attribs[self::NS_XML . ' lang'] ) |
| 1231 | || !preg_match( '/^[-A-Za-z0-9]{2,}$/D', $attribs[self::NS_XML . ' lang'] ) |
| 1232 | ) { |
| 1233 | throw new RuntimeException( __METHOD__ |
| 1234 | . " <rdf:li> did not contain, or has invalid xml:lang attribute in lang alternative" ); |
| 1235 | } |
| 1236 | |
| 1237 | // Lang is case-insensitive. |
| 1238 | $this->itemLang = strtolower( $attribs[self::NS_XML . ' lang'] ); |
| 1239 | |
| 1240 | // need to add curItem[0] on again since one is for the specific item |
| 1241 | // and one is for the entire group. |
| 1242 | array_unshift( $this->curItem, $this->curItem[0] ); |
| 1243 | array_unshift( $this->mode, self::MODE_SIMPLE ); |
| 1244 | $this->processingArray = true; |
| 1245 | } |
| 1246 | |
| 1247 | /** |
| 1248 | * Hits an opening element. |
| 1249 | * Generally just calls a helper based on what MODE we're in. |
| 1250 | * Also does some initial set up for the wrapper element |
| 1251 | * |
| 1252 | * @param resource $parser |
| 1253 | * @param string $elm Namespace "<space>" element |
| 1254 | * @param array $attribs Attribute name => value |
| 1255 | * @throws RuntimeException |
| 1256 | */ |
| 1257 | public function startElement( $parser, $elm, $attribs ): void { |
| 1258 | if ( $elm === self::NS_RDF . ' RDF' |
| 1259 | || $elm === 'adobe:ns:meta/ xmpmeta' |
| 1260 | || $elm === 'adobe:ns:meta/ xapmeta' |
| 1261 | ) { |
| 1262 | /* ignore. */ |
| 1263 | return; |
| 1264 | } |
| 1265 | |
| 1266 | if ( $elm === self::NS_RDF . ' Description' ) { |
| 1267 | if ( count( $this->mode ) === 0 ) { |
| 1268 | // outer rdf:desc |
| 1269 | array_unshift( $this->mode, self::MODE_INITIAL ); |
| 1270 | } |
| 1271 | } elseif ( $elm === self::NS_RDF . ' type' ) { |
| 1272 | // This doesn't support rdf:type properly. |
| 1273 | // In practise, I have yet to see a file that |
| 1274 | // uses this element, however it is mentioned |
| 1275 | // on page 25 of part 1 of the xmp standard. |
| 1276 | // Also, it seems as if exiv2 and exiftool do not support |
| 1277 | // this either (That or I misunderstand the standard) |
| 1278 | $this->logger->info( |
| 1279 | __METHOD__ . ' Encountered <rdf:type> which isn\'t currently supported', |
| 1280 | [ 'file' => $this->filename ] |
| 1281 | ); |
| 1282 | } |
| 1283 | |
| 1284 | if ( strpos( $elm, ' ' ) === false ) { |
| 1285 | // This probably shouldn't happen. |
| 1286 | $this->logger->info( |
| 1287 | __METHOD__ . " Encountered <$elm> which has no namespace. Skipping.", |
| 1288 | [ 'file' => $this->filename ] |
| 1289 | ); |
| 1290 | |
| 1291 | return; |
| 1292 | } |
| 1293 | |
| 1294 | [ $ns, $tag ] = explode( ' ', $elm, 2 ); |
| 1295 | |
| 1296 | if ( count( $this->mode ) === 0 ) { |
| 1297 | // This should not happen. |
| 1298 | throw new RuntimeException( 'Error extracting XMP, ' |
| 1299 | . "encountered <$elm> with no mode" ); |
| 1300 | } |
| 1301 | |
| 1302 | switch ( $this->mode[0] ) { |
| 1303 | case self::MODE_IGNORE: |
| 1304 | $this->startElementModeIgnore( $elm ); |
| 1305 | break; |
| 1306 | case self::MODE_SIMPLE: |
| 1307 | $this->startElementModeSimple( $elm, $attribs ); |
| 1308 | break; |
| 1309 | case self::MODE_INITIAL: |
| 1310 | $this->startElementModeInitial( $ns, $tag, $attribs ); |
| 1311 | break; |
| 1312 | case self::MODE_STRUCT: |
| 1313 | $this->startElementModeStruct( $ns, $tag, $attribs ); |
| 1314 | break; |
| 1315 | case self::MODE_BAG: |
| 1316 | case self::MODE_BAGSTRUCT: |
| 1317 | $this->startElementModeBag( $elm ); |
| 1318 | break; |
| 1319 | case self::MODE_SEQ: |
| 1320 | $this->startElementModeSeq( $elm ); |
| 1321 | break; |
| 1322 | case self::MODE_LANG: |
| 1323 | $this->startElementModeLang( $elm ); |
| 1324 | break; |
| 1325 | case self::MODE_LI_LANG: |
| 1326 | $this->startElementModeLiLang( $elm, $attribs ); |
| 1327 | break; |
| 1328 | case self::MODE_LI: |
| 1329 | $this->startElementModeLi( $elm, $attribs ); |
| 1330 | break; |
| 1331 | case self::MODE_QDESC: |
| 1332 | $this->startElementModeQDesc( $elm ); |
| 1333 | break; |
| 1334 | default: |
| 1335 | throw new RuntimeException( 'StartElement in unknown mode: ' . $this->mode[0] ); |
| 1336 | } |
| 1337 | } |
| 1338 | |
| 1339 | /** |
| 1340 | * Process attributes. |
| 1341 | * Simple values can be stored as either a tag or attribute |
| 1342 | * |
| 1343 | * Often the initial "<rdf:Description>" tag just has all the simple |
| 1344 | * properties as attributes. |
| 1345 | * |
| 1346 | * @par Example: |
| 1347 | * @code |
| 1348 | * <rdf:Description rdf:about="" |
| 1349 | * xmlns:exif="http://ns.adobe.com/exif/1.0/" exif:DigitalZoomRatio="0/10"> |
| 1350 | * @endcode |
| 1351 | * |
| 1352 | * @param array $attribs Array attribute=>value |
| 1353 | * @throws RuntimeException |
| 1354 | */ |
| 1355 | private function doAttribs( $attribs ): void { |
| 1356 | // first check for rdf:parseType attribute, as that can change |
| 1357 | // how the attributes are interpreted. |
| 1358 | |
| 1359 | if ( isset( $attribs[self::NS_RDF . ' parseType'] ) |
| 1360 | && $attribs[self::NS_RDF . ' parseType'] === 'Resource' |
| 1361 | && $this->mode[0] === self::MODE_SIMPLE |
| 1362 | ) { |
| 1363 | // this is equivalent to having an inner rdf:Description |
| 1364 | $this->mode[0] = self::MODE_QDESC; |
| 1365 | } |
| 1366 | foreach ( $attribs as $name => $val ) { |
| 1367 | if ( strpos( $name, ' ' ) === false ) { |
| 1368 | // This shouldn't happen, but so far some old software forgets namespace |
| 1369 | // on rdf:about. |
| 1370 | $this->logger->info( |
| 1371 | __METHOD__ . ' Encountered non-namespaced attribute: ' . |
| 1372 | " $name=\"$val\". Skipping. ", |
| 1373 | [ 'file' => $this->filename ] |
| 1374 | ); |
| 1375 | continue; |
| 1376 | } |
| 1377 | [ $ns, $tag ] = explode( ' ', $name, 2 ); |
| 1378 | if ( $ns === self::NS_RDF ) { |
| 1379 | if ( $tag === 'value' || $tag === 'resource' ) { |
| 1380 | // resource is for url. |
| 1381 | // value attribute is a weird way of just putting the contents. |
| 1382 | $this->char( $this->xmlParser, $val ); |
| 1383 | } |
| 1384 | } elseif ( isset( $this->items[$ns][$tag] ) ) { |
| 1385 | if ( $this->mode[0] === self::MODE_SIMPLE ) { |
| 1386 | throw new RuntimeException( __METHOD__ |
| 1387 | . " $ns:$tag found as attribute where not allowed" ); |
| 1388 | } |
| 1389 | $this->saveValue( $ns, $tag, $val ); |
| 1390 | } else { |
| 1391 | $this->logger->debug( |
| 1392 | __METHOD__ . " Ignoring unrecognized element <$ns:$tag>.", |
| 1393 | [ 'file' => $this->filename ] |
| 1394 | ); |
| 1395 | } |
| 1396 | } |
| 1397 | } |
| 1398 | |
| 1399 | /** |
| 1400 | * Given an extracted value, save it to results array |
| 1401 | * |
| 1402 | * note also uses $this->ancestorStruct and |
| 1403 | * $this->processingArray to determine what name to |
| 1404 | * save the value under. (in addition to $tag). |
| 1405 | * |
| 1406 | * @param string $ns Namespace of tag this is for |
| 1407 | * @param string $tag Tag name |
| 1408 | * @param string $val Value to save |
| 1409 | */ |
| 1410 | private function saveValue( $ns, $tag, $val ): void { |
| 1411 | $info =& $this->items[$ns][$tag]; |
| 1412 | $finalName = $info['map_name'] ?? $tag; |
| 1413 | if ( isset( $info['validate'] ) ) { |
| 1414 | if ( is_array( $info['validate'] ) ) { |
| 1415 | $validate = $info['validate']; |
| 1416 | } else { |
| 1417 | $validator = new Validate( $this->logger ); |
| 1418 | $validate = [ $validator, $info['validate'] ]; |
| 1419 | } |
| 1420 | |
| 1421 | if ( is_callable( $validate ) ) { |
| 1422 | call_user_func_array( $validate, [ $info, &$val, true ] ); |
| 1423 | // the reasoning behind using &$val instead of using the return value |
| 1424 | // is to be consistent between here and validating structures. |
| 1425 | if ( $val === null ) { |
| 1426 | $this->logger->info( |
| 1427 | __METHOD__ . " <$ns:$tag> failed validation.", |
| 1428 | [ 'file' => $this->filename ] |
| 1429 | ); |
| 1430 | |
| 1431 | return; |
| 1432 | } |
| 1433 | } else { |
| 1434 | $this->logger->warning( |
| 1435 | __METHOD__ . " Validation function for $finalName (" . |
| 1436 | get_class( $validate[0] ) . '::' . $validate[1] . '()) is not callable.', |
| 1437 | [ 'file' => $this->filename ] |
| 1438 | ); |
| 1439 | } |
| 1440 | } |
| 1441 | |
| 1442 | if ( $this->ancestorStruct && $this->processingArray ) { |
| 1443 | // Aka both an array and a struct. ( self::MODE_BAGSTRUCT ) |
| 1444 | $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][][$finalName] = $val; |
| 1445 | } elseif ( $this->ancestorStruct ) { |
| 1446 | $this->results['xmp-' . $info['map_group']][$this->ancestorStruct][$finalName] = $val; |
| 1447 | } elseif ( $this->processingArray ) { |
| 1448 | if ( $this->itemLang === false ) { |
| 1449 | // normal array |
| 1450 | $this->results['xmp-' . $info['map_group']][$finalName][] = $val; |
| 1451 | } else { |
| 1452 | // lang array. |
| 1453 | $this->results['xmp-' . $info['map_group']][$finalName][$this->itemLang] = $val; |
| 1454 | } |
| 1455 | } else { |
| 1456 | $this->results['xmp-' . $info['map_group']][$finalName] = $val; |
| 1457 | } |
| 1458 | } |
| 1459 | } |