| Index: branches/img_metadata/phase3/maintenance/language/messages.inc |
| — | — | @@ -2604,7 +2604,6 @@ |
| 2605 | 2605 | 'exif-stripbytecounts', |
| 2606 | 2606 | 'exif-jpeginterchangeformat', |
| 2607 | 2607 | 'exif-jpeginterchangeformatlength', |
| 2608 | | - 'exif-transferfunction', |
| 2609 | 2608 | 'exif-whitepoint', |
| 2610 | 2609 | 'exif-primarychromaticities', |
| 2611 | 2610 | 'exif-ycbcrcoefficients', |
| — | — | @@ -2734,6 +2733,9 @@ |
| 2735 | 2734 | 'exif-datetimereleased', |
| 2736 | 2735 | 'exif-originaltransmissionref', |
| 2737 | 2736 | 'exif-identifier', |
| | 2737 | + 'exif-lens', |
| | 2738 | + 'exif-serialnumber', |
| | 2739 | + |
| 2738 | 2740 | ), |
| 2739 | 2741 | 'exif-values' => array( |
| 2740 | 2742 | 'exif-make-value', |
| Index: branches/img_metadata/phase3/maintenance/language/messageTypes.inc |
| — | — | @@ -386,7 +386,6 @@ |
| 387 | 387 | 'exif-stripbytecounts', |
| 388 | 388 | 'exif-jpeginterchangeformat', |
| 389 | 389 | 'exif-jpeginterchangeformatlength', |
| 390 | | - 'exif-transferfunction', |
| 391 | 390 | 'exif-whitepoint', |
| 392 | 391 | 'exif-primarychromaticities', |
| 393 | 392 | 'exif-ycbcrcoefficients', |
| — | — | @@ -647,4 +646,7 @@ |
| 648 | 647 | 'exif-dc-rights', |
| 649 | 648 | 'exif-dc-source', |
| 650 | 649 | 'exif-dc-type', |
| | 650 | + 'exif-lens', |
| | 651 | + 'exif-serialnumber', |
| | 652 | + |
| 651 | 653 | ); |
| Index: branches/img_metadata/phase3/includes/Exif.php |
| — | — | @@ -139,7 +139,7 @@ |
| 140 | 140 | 'JPEGInterchangeFormatLength' => Exif::SHORT.','.Exif::LONG, # Bytes of JPEG data |
| 141 | 141 | |
| 142 | 142 | # Tags relating to image data characteristics |
| 143 | | - 'TransferFunction' => Exif::SHORT, # Transfer function |
| | 143 | + 'TransferFunction' => Exif::IGNORE, # Transfer function |
| 144 | 144 | 'WhitePoint' => array( Exif::RATIONAL, 2), # White point chromaticity |
| 145 | 145 | 'PrimaryChromaticities' => array( Exif::RATIONAL, 6), # Chromaticities of primarities |
| 146 | 146 | 'YCbCrCoefficients' => array( Exif::RATIONAL, 3), # Color space transformation matrix coefficients #p27 |
| — | — | @@ -1382,6 +1382,8 @@ |
| 1383 | 1383 | case 'dc-rights': |
| 1384 | 1384 | case 'dc-source': |
| 1385 | 1385 | case 'dc-type': |
| | 1386 | + case 'Lens': |
| | 1387 | + case 'SerialNumber': |
| 1386 | 1388 | |
| 1387 | 1389 | $val = htmlspecialchars( $val ); |
| 1388 | 1390 | break; |
| Index: branches/img_metadata/phase3/includes/media/BitmapMetadataHandler.php |
| — | — | @@ -3,8 +3,9 @@ |
| 4 | 4 | Class to deal with reconciling and extracting metadata from bitmap images. |
| 5 | 5 | This is meant to comply with http://www.metadataworkinggroup.org/pdf/mwg_guidance.pdf |
| 6 | 6 | |
| 7 | | -@todo finish IPTC |
| 8 | | -@todo xmp |
| | 7 | +This sort of acts as an intermediary between MediaHandler::getMetadata |
| | 8 | +and the various metadata extractors. |
| | 9 | + |
| 9 | 10 | @todo other image formats. |
| 10 | 11 | */ |
| 11 | 12 | class BitmapMetadataHandler { |
| — | — | @@ -12,118 +13,21 @@ |
| 13 | 14 | // the max segment is a sanity check. |
| 14 | 15 | // A jpeg file should never even remotely have |
| 15 | 16 | // that many segments. Your average file has about 10. |
| 16 | | - private $filetype; |
| 17 | 17 | private $filename; |
| 18 | 18 | private $metadata = Array(); |
| 19 | 19 | private $metaPriority = Array( |
| 20 | 20 | 20 => Array( 'other' ), |
| 21 | 21 | 40 => Array( 'file-comment' ), |
| 22 | | - 60 => Array( 'iptc-bad-hash' ), |
| | 22 | + 60 => Array( 'iptc-good-hash', 'iptc-no-hash' ), |
| 23 | 23 | 70 => Array( 'xmp-deprected' ), |
| 24 | 24 | 80 => Array( 'xmp-general' ), |
| 25 | 25 | 90 => Array( 'xmp-exif' ), |
| 26 | | - 100 => Array( 'iptc-good-hash', 'iptc-no-hash' ), |
| | 26 | + 100 => Array( 'iptc-bad-hash' ), |
| 27 | 27 | 120 => Array( 'exif' ), |
| 28 | 28 | ); |
| 29 | 29 | private $iptcType = 'iptc-no-hash'; |
| 30 | 30 | |
| 31 | | - /** Function to extract metadata segmants of interest from jpeg files |
| 32 | | - * based on GIFMetadataExtractor. |
| 33 | | - * |
| 34 | | - * we can almost use getimagesize to do this |
| 35 | | - * but gis doesn't support having multiple app1 segments |
| 36 | | - * and those can't extract xmp on files containing both exif and xmp data |
| 37 | | - * |
| 38 | | - * I'm not sure if this should be in a class of its own, like GIFMetadataExtractor |
| 39 | | - * |
| 40 | | - * @param String $filename name of jpeg file |
| 41 | | - * @return Array of interesting segments. |
| 42 | | - * Can throw an exception if given invalid file. |
| 43 | | - */ |
| 44 | | - function jpegSegmentSplitter () { |
| 45 | | - $filename = $this->filename; |
| 46 | | - $segmentCount = 0; |
| 47 | | - if ( $this->filetype !== 'image/jpeg' ) throw new MWException( "jpegSegmentSplitter called on non-jpeg" ); |
| 48 | | - |
| 49 | | - $segments = Array( 'XMP_ext' => array(), 'COM' => array() ); |
| 50 | | - |
| 51 | | - if ( !$filename ) throw new MWException( "No filename specified for BitmapMetadataHandler" ); |
| 52 | | - if ( !file_exists( $filename ) || is_dir( $filename ) ) throw new MWException( "Invalid file $filename passed to BitmapMetadataHandler" ); |
| 53 | | - |
| 54 | | - $fh = fopen( $filename, "rb" ); |
| 55 | | - |
| 56 | | - if ( !$fh ) throw new MWException( "Could not open file $filename" ); |
| 57 | | - |
| 58 | | - $buffer = fread( $fh, 2 ); |
| 59 | | - if ( $buffer !== "\xFF\xD8" ) throw new MWException( "Not a jpeg, no SOI" ); |
| 60 | | - while ( !feof( $fh ) ) { |
| 61 | | - $buffer = fread( $fh, 1 ); |
| 62 | | - $segmentCount++; |
| 63 | | - if ( $segmentCount > self::MAX_JPEG_SEGMENTS ) { |
| 64 | | - // this is just a sanity check |
| 65 | | - throw new MWException('Too many jpeg segments. Aborting'); |
| 66 | | - } |
| 67 | | - if ( $buffer !== "\xFF" ) { |
| 68 | | - throw new MWException( "Error reading jpeg file marker" ); |
| 69 | | - } |
| 70 | | - |
| 71 | | - $buffer = fread( $fh, 1 ); |
| 72 | | - if ( $buffer === "\xFE" ) { |
| 73 | | - // COM section -- file comment |
| 74 | | - // First see if valid utf-8, |
| 75 | | - // if not try to convert it to windows-1252. |
| 76 | | - $com = $oldCom = trim( self::jpegExtractMarker( $fh ) ); |
| 77 | | - UtfNormal::quickIsNFCVerify( $com ); |
| 78 | | - //turns $com to valid utf-8. |
| 79 | | - //thus if no change, its utf-8, otherwise its something else. |
| 80 | | - if ( $com !== $oldCom ) { |
| 81 | | - $oldCom = iconv( 'windows-1252', 'UTF-8//IGNORE', $oldCom ); |
| 82 | | - } |
| 83 | | - $segments["COM"][] = $oldCom; |
| 84 | | - |
| 85 | | - } elseif ( $buffer === "\xE1" ) { |
| 86 | | - // APP1 section (Exif, XMP, and XMP extended) |
| 87 | | - $temp = self::jpegExtractMarker( $fh ); |
| 88 | | - |
| 89 | | - // check what type of app segment this is. |
| 90 | | - if ( substr( $temp, 0, 29 ) === "http://ns.adobe.com/xap/1.0/\x00" ) { |
| 91 | | - $segments["XMP"] = substr( $temp, 29 ); |
| 92 | | - } elseif ( substr( $temp, 0, 35 ) === "http://ns.adobe.com/xmp/extension/\x00" ) { |
| 93 | | - $segments["XMP_ext"][] = substr( $temp, 35 ); |
| 94 | | - } |
| 95 | | - } elseif ( $buffer === "\xED" ) { |
| 96 | | - // APP13 - PSIR. IPTC and some photoshop stuff |
| 97 | | - $temp = self::jpegExtractMarker( $fh ); |
| 98 | | - if ( substr( $temp, 0, 14 ) === "Photoshop 3.0\x00" ) { |
| 99 | | - $segments["PSIR"] = $temp; |
| 100 | | - } |
| 101 | | - } elseif ( $buffer === "\xD9" || $buffer === "\xDA" ) { |
| 102 | | - // EOI - end of image or SOS - start of scan. either way we're past any interesting segments |
| 103 | | - return $segments; |
| 104 | | - } else { |
| 105 | | - // segment we don't care about, so skip |
| 106 | | - $size = unpack( "nint", fread( $fh, 2 ) ); |
| 107 | | - if ( $size['int'] <= 2 ) throw new MWException( "invalid marker size in jpeg" ); |
| 108 | | - fseek( $fh, $size['int'] - 2, SEEK_CUR ); |
| 109 | | - } |
| 110 | | - |
| 111 | | - } |
| 112 | | - // shouldn't get here. |
| 113 | | - throw new MWException( "Reached end of jpeg file unexpectedly" ); |
| 114 | | - |
| 115 | | - } |
| 116 | 31 | /** |
| 117 | | - * Helper function for jpegSegmentSplitter |
| 118 | | - * @param &$fh FileHandle for jpeg file |
| 119 | | - * @return data content of segment. |
| 120 | | - */ |
| 121 | | - private function jpegExtractMarker( &$fh ) { |
| 122 | | - $size = unpack( "nint", fread( $fh, 2 ) ); |
| 123 | | - if ( $size['int'] <= 2 ) throw new MWException( "invalid marker size in jpeg" ); |
| 124 | | - return fread( $fh, $size['int'] - 2 ); |
| 125 | | - } |
| 126 | | - |
| 127 | | - /** |
| 128 | 32 | * This does the photoshop image resource app13 block |
| 129 | 33 | * of interest, IPTC-IIM metadata is stored here. |
| 130 | 34 | * |
| — | — | @@ -132,107 +36,13 @@ |
| 133 | 37 | * @param String $app13 String containing app13 block from jpeg file |
| 134 | 38 | */ |
| 135 | 39 | private function doApp13 ( $app13 ) { |
| 136 | | - $this->doPSIR( $app13 ); |
| | 40 | + $this->iptcType = JpegMetadataExtractor::doPSIR( $app13 ); |
| 137 | 41 | |
| 138 | 42 | $iptc = IPTC::parse( $app13 ); |
| 139 | 43 | $this->addMetadata( $iptc, $this->iptcType ); |
| 140 | 44 | } |
| 141 | 45 | |
| 142 | | - /** |
| 143 | | - * This reads the photoshop image resource. |
| 144 | | - * Currently it only compares the iptc/iim hash |
| 145 | | - * with the stored hash, which is used to determine the precedence |
| 146 | | - * of the iptc data. In future it may extract some other info, like |
| 147 | | - * url of copyright license. |
| 148 | | - * |
| 149 | | - * This should generally be called by doApp13() |
| 150 | | - * |
| 151 | | - * @param String $app13 photoshop psir app13 block from jpg. |
| 152 | | - */ |
| 153 | | - private function doPSIR ( $app13 ) { |
| 154 | | - if ( !$app13 ) return; |
| 155 | | - // First compare hash with real thing |
| 156 | | - // 0x404 contains IPTC, 0x425 has hash |
| 157 | | - // This is used to determine if the iptc is newer than |
| 158 | | - // the xmp data, as xmp programs update the hash, |
| 159 | | - // where non-xmp programs don't. |
| 160 | 46 | |
| 161 | | - $offset = 14; // skip past PHOTOSHOP 3.0 identifier. should already be checked. |
| 162 | | - $appLen = strlen( $app13 ); |
| 163 | | - $realHash = ""; |
| 164 | | - $recordedHash = ""; |
| 165 | | - |
| 166 | | - // the +12 is the length of an empty item. |
| 167 | | - while ( $offset + 12 <= $appLen ) { |
| 168 | | - $valid = true; |
| 169 | | - $id = false; |
| 170 | | - $lenName = false; |
| 171 | | - $lenData = false; |
| 172 | | - |
| 173 | | - if ( substr( $app13, $offset, 4 ) !== '8BIM' ) { |
| 174 | | - // its supposed to be 8BIM |
| 175 | | - // but apperently sometimes isn't esp. in |
| 176 | | - // really old jpg's |
| 177 | | - $valid = false; |
| 178 | | - } |
| 179 | | - $offset += 4; |
| 180 | | - $id = substr( $app13, $offset, 2 ); |
| 181 | | - // id is a 2 byte id number which identifies |
| 182 | | - // the piece of info this record contains. |
| 183 | | - |
| 184 | | - $offset += 2; |
| 185 | | - |
| 186 | | - // some record types can contain a name, which |
| 187 | | - // is a pascal string 0-padded to be an even |
| 188 | | - // number of bytes. Most times (and any time |
| 189 | | - // we care) this is empty, making it two null bytes. |
| 190 | | - |
| 191 | | - $lenName = ord( substr( $app13, $offset, 1 ) ) + 1; |
| 192 | | - // we never use the name so skip it. +1 for length byte |
| 193 | | - if ( $lenName % 2 == 1 ) $lenName++; // pad to even. |
| 194 | | - $offset += $lenName; |
| 195 | | - |
| 196 | | - // now length of data (unsigned long big endian) |
| 197 | | - $lenData = unpack( 'Nlen', substr( $app13, $offset, 4 ) ); |
| 198 | | - $offset += 4; // 4bytes length field; |
| 199 | | - |
| 200 | | - // this should not happen, but check. |
| 201 | | - if ( $lenData['len'] + $offset > $appLen ) { |
| 202 | | - wfDebug( __METHOD__ . ' PSIR data too long.' ); |
| 203 | | - return false; |
| 204 | | - } |
| 205 | | - |
| 206 | | - if ( $valid ) { |
| 207 | | - switch ( $id ) { |
| 208 | | - case "\x04\x04": |
| 209 | | - // IPTC block |
| 210 | | - $realHash = md5( substr( $app13, $offset, $lenData['len'] ), true ); |
| 211 | | - break; |
| 212 | | - case "\x04\x25": |
| 213 | | - $recordedHash = substr( $app13, $offset, $lenData['len'] ); |
| 214 | | - break; |
| 215 | | - } |
| 216 | | - } |
| 217 | | - |
| 218 | | - // if odd, add 1 to length to account for |
| 219 | | - // null pad byte. |
| 220 | | - if ( $lenData['len'] % 2 == 1 ) $lenData['len']++; |
| 221 | | - $offset += $lenData['len']; |
| 222 | | - |
| 223 | | - } |
| 224 | | - |
| 225 | | - if ( !$realHash ) return false; // no iptc data |
| 226 | | - |
| 227 | | - if ( !$recordedHash ) { |
| 228 | | - $this->iptcType = 'iptc-no-hash'; |
| 229 | | - } elseif ( $realHash === $recordedHash ) { |
| 230 | | - $this->iptcType = 'iptc-good-hash'; |
| 231 | | - } else { /*$realHash !== $recordedHash */ |
| 232 | | - $this->iptcType = 'iptc-bad-hash'; |
| 233 | | - } |
| 234 | | - |
| 235 | | - } |
| 236 | | - |
| 237 | 47 | /** get exif info using exif class. |
| 238 | 48 | * Basically what used to be in BitmapHandler::getMetadata(). |
| 239 | 49 | * Just calls stuff in the Exif class. |
| — | — | @@ -300,34 +110,32 @@ |
| 301 | 111 | |
| 302 | 112 | /** constructor. |
| 303 | 113 | * This generally shouldn't be called directly |
| 304 | | - * instead BitmapMetadataHandler::newForJpeg should be used. |
| | 114 | + * instead one of the static methods should be used |
| 305 | 115 | * |
| 306 | 116 | * @param string $file - full path to file |
| 307 | | - * @param string $type - mime type of file |
| 308 | 117 | */ |
| 309 | | - function __construct ( $file, $type ) { |
| | 118 | + function __construct ( $file ) { |
| 310 | 119 | $this->filename = $file; |
| 311 | | - $this->filetype = $type; |
| 312 | 120 | } |
| 313 | | - /** factory function special for jpg's. |
| 314 | | - * This is how new BitmapMetadataHandler's should be made. |
| 315 | | - * at some point there should be a newForPNG, etc. |
| | 121 | + /** Main entry point for jpeg's. |
| 316 | 122 | * |
| 317 | 123 | * @param string $file filename (with full path) |
| 318 | | - * @return BitmapMetadataHandler |
| | 124 | + * @return metadata result array. |
| | 125 | + * @throws MWException on invalid file. |
| 319 | 126 | */ |
| 320 | | - static function newForJpeg ( $file ) { |
| 321 | | - $meta = new self( $file, 'image/jpeg' ); |
| | 127 | + static function Jpeg ( $file ) { |
| | 128 | + global $wgShowXMP; |
| | 129 | + $meta = new self( $file ); |
| 322 | 130 | $meta->getExif(); |
| 323 | 131 | $seg = Array(); |
| 324 | | - $seg = $meta->JpegSegmentSplitter(); |
| | 132 | + $seg = JpegMetadataExtractor::segmentSplitter( $file ); |
| 325 | 133 | if ( isset( $seg['COM'] ) && isset( $seg['COM'][0] ) ) { |
| 326 | 134 | $meta->addMetadata( Array( 'JPEGFileComment' => $seg['COM'] ), 'file-comment' ); |
| 327 | 135 | } |
| 328 | 136 | if ( isset( $seg['PSIR'] ) ) { |
| 329 | 137 | $meta->doApp13( $seg['PSIR'] ); |
| 330 | 138 | } |
| 331 | | - if ( isset( $seg['XMP'] ) ) { |
| | 139 | + if ( isset( $seg['XMP'] ) && $wgShowXMP ) { |
| 332 | 140 | $xmp = new XMPReader(); |
| 333 | 141 | $xmp->parse( $seg['XMP'] ); |
| 334 | 142 | foreach( $seg['XMP_ext'] as $xmpExt ) { |
| — | — | @@ -342,7 +150,33 @@ |
| 343 | 151 | $meta->addMetadata( $array, $type ); |
| 344 | 152 | } |
| 345 | 153 | } |
| 346 | | - return $meta; |
| | 154 | + return $meta->getMetadataArray(); |
| 347 | 155 | } |
| | 156 | + /** Entry point for png |
| | 157 | + * At some point in the future this might |
| | 158 | + * merge the png various tEXt chunks to that |
| | 159 | + * are interesting, but for now it only does XMP |
| | 160 | + * |
| | 161 | + * @param $filename String full path to file |
| | 162 | + * @return Array Array for storage in img_metadata. |
| | 163 | + */ |
| | 164 | + static public function PNG( $filename ) { |
| | 165 | + global $wgShowXMP; |
| 348 | 166 | |
| | 167 | + $meta = new self( $filename ); |
| | 168 | + $array = PNGMetadataExtractor::getMetadata( $filename ); |
| | 169 | + if ( isset( $array['xmp'] ) && $array['xmp'] !== '' && $wgShowXMP) { |
| | 170 | + $xmp = new XMPReader(); |
| | 171 | + $xmp->parse($array['xmp']); |
| | 172 | + $xmpRes = $xmp->getResults(); |
| | 173 | + foreach( $xmpRes as $type => $xmpSection ) { |
| | 174 | + $meta->addMetadata( $xmpSection, $type ); |
| | 175 | + } |
| | 176 | + } |
| | 177 | + unset( $array['xmp'] ); |
| | 178 | + $array['metadata'] = $meta->getMetadataArray(); |
| | 179 | + $array['metadata']['_MW_PNG_VERSION'] = '1'; |
| | 180 | + return $array; |
| | 181 | + } |
| | 182 | + |
| 349 | 183 | } |
| Index: branches/img_metadata/phase3/includes/media/Bitmap.php |
| — | — | @@ -413,10 +413,6 @@ |
| 414 | 414 | } |
| 415 | 415 | |
| 416 | 416 | function formatMetadata( $image ) { |
| 417 | | - $result = array( |
| 418 | | - 'visible' => array(), |
| 419 | | - 'collapsed' => array() |
| 420 | | - ); |
| 421 | 417 | $metadata = $image->getMetadata(); |
| 422 | 418 | if ( !$metadata ) { |
| 423 | 419 | return false; |
| — | — | @@ -426,20 +422,7 @@ |
| 427 | 423 | return false; |
| 428 | 424 | } |
| 429 | 425 | unset( $exif['MEDIAWIKI_EXIF_VERSION'] ); |
| 430 | | - $format = new FormatExif( $exif ); |
| 431 | | - |
| 432 | | - $formatted = $format->getFormattedData(); |
| 433 | | - // Sort fields into visible and collapsed |
| 434 | | - $visibleFields = $this->visibleMetadataFields(); |
| 435 | | - foreach ( $formatted as $name => $value ) { |
| 436 | | - $tag = strtolower( $name ); |
| 437 | | - self::addMeta( $result, |
| 438 | | - in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed', |
| 439 | | - 'exif', |
| 440 | | - $tag, |
| 441 | | - $value |
| 442 | | - ); |
| 443 | | - } |
| 444 | | - return $result; |
| | 426 | + return $this->formatMetadataHelper( $exif ); |
| 445 | 427 | } |
| | 428 | + |
| 446 | 429 | } |
| Index: branches/img_metadata/phase3/includes/media/Generic.php |
| — | — | @@ -266,6 +266,37 @@ |
| 267 | 267 | return false; |
| 268 | 268 | } |
| 269 | 269 | |
| | 270 | + /** sorts the visible/invisible field. |
| | 271 | + * Split off from ImageHandler::formatMetadata, as used by more than |
| | 272 | + * one type of handler. |
| | 273 | + * |
| | 274 | + * This is used by the media handlers that use the FormatExif class |
| | 275 | + * |
| | 276 | + * @param $metadataArray Array metadata array |
| | 277 | + * @return array for use displaying metadata. |
| | 278 | + */ |
| | 279 | + function formatMetadataHelper( $metadataArray ) { |
| | 280 | + $result = array( |
| | 281 | + 'visible' => array(), |
| | 282 | + 'collapsed' => array() |
| | 283 | + ); |
| | 284 | + $format = new FormatExif( $metadataArray ); |
| | 285 | + |
| | 286 | + $formatted = $format->getFormattedData(); |
| | 287 | + // Sort fields into visible and collapsed |
| | 288 | + $visibleFields = $this->visibleMetadataFields(); |
| | 289 | + foreach ( $formatted as $name => $value ) { |
| | 290 | + $tag = strtolower( $name ); |
| | 291 | + self::addMeta( $result, |
| | 292 | + in_array( $tag, $visibleFields ) ? 'visible' : 'collapsed', |
| | 293 | + 'exif', |
| | 294 | + $tag, |
| | 295 | + $value |
| | 296 | + ); |
| | 297 | + } |
| | 298 | + return $result; |
| | 299 | + } |
| | 300 | + |
| 270 | 301 | /** |
| 271 | 302 | * @todo Fixme: document this! |
| 272 | 303 | * 'value' thingy goes into a wikitext table; it used to be escaped but |
| Index: branches/img_metadata/phase3/includes/media/XMPValidate.php |
| — | — | @@ -49,13 +49,31 @@ |
| 50 | 50 | // this only validates standalone properties, not arrays, etc |
| 51 | 51 | return; |
| 52 | 52 | } |
| 53 | | - if ( !preg_match( '/^(-?\d+)\/(\d+[1-9]|[1-9]\d*)$/', $val ) ) { |
| | 53 | + if ( !preg_match( '/^(?:-?\d+)\/(?:\d+[1-9]|[1-9]\d*)$/D', $val ) ) { |
| 54 | 54 | wfDebugLog( 'XMP', __METHOD__ . " Expected rational but got $val" ); |
| 55 | 55 | $val = null; |
| 56 | 56 | } |
| 57 | 57 | |
| 58 | 58 | } |
| 59 | 59 | /** |
| | 60 | + * function to validate integers |
| | 61 | + * |
| | 62 | + * @param $info Array information about current property |
| | 63 | + * @param &$val Mixed current value to validate |
| | 64 | + * @param $standalone Boolean if this is a simple property or array |
| | 65 | + */ |
| | 66 | + public static function validateInteger( $info, &$val, $standalone ) { |
| | 67 | + if ( !$standalone ) { |
| | 68 | + // this only validates standalone properties, not arrays, etc |
| | 69 | + return; |
| | 70 | + } |
| | 71 | + if ( !preg_match( '/^[-+]?\d+$/D', $val ) ) { |
| | 72 | + wfDebugLog( 'XMP', __METHOD__ . " Expected integer but got $val" ); |
| | 73 | + $val = null; |
| | 74 | + } |
| | 75 | + |
| | 76 | + } |
| | 77 | + /** |
| 60 | 78 | * function to validate properties with a fixed number of allowed |
| 61 | 79 | * choices. (closed choice) |
| 62 | 80 | * |
| Index: branches/img_metadata/phase3/includes/media/XMPInfo.php |
| — | — | @@ -188,6 +188,142 @@ |
| 189 | 189 | 'mode' => XMPReader::MODE_SEQ, |
| 190 | 190 | ), |
| 191 | 191 | ), |
| | 192 | + 'http://ns.adobe.com/tiff/1.0/' => array( |
| | 193 | + 'Artist' => array( |
| | 194 | + 'map_group' => 'exif', |
| | 195 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 196 | + ), |
| | 197 | + 'BitsPerSample' => array( |
| | 198 | + 'map_group' => 'exif', |
| | 199 | + 'mode' => XMPReader::MODE_SEQ, |
| | 200 | + 'validate' => 'validateInteger', |
| | 201 | + ), |
| | 202 | + 'Compression' => array( |
| | 203 | + 'map_group' => 'exif', |
| | 204 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 205 | + 'validate' => 'validateClosed', |
| | 206 | + 'choices' => array( '1' => true, '6' => true ), |
| | 207 | + ), |
| | 208 | + /* this prop should not be used in XMP. dc:rights is the correct prop */ |
| | 209 | + 'Copyright' => array( |
| | 210 | + 'map_group' => 'exif', |
| | 211 | + 'mode' => XMPReader::MODE_LANG, |
| | 212 | + ), |
| | 213 | + 'DateTime' => array( /* proper prop is xmp:ModifyDate */ |
| | 214 | + 'map_group' => 'exif', |
| | 215 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 216 | + 'validate' => 'validateDate', |
| | 217 | + ), |
| | 218 | + 'ImageDescription' => array( /* proper one is dc:description */ |
| | 219 | + 'map_group' => 'exif', |
| | 220 | + 'mode' => XMPReader::MODE_LANG, |
| | 221 | + ), |
| | 222 | + 'ImageLength' => array( |
| | 223 | + 'map_group' => 'exif', |
| | 224 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 225 | + 'validate' => 'validateInteger', |
| | 226 | + ), |
| | 227 | + 'ImageWidth' => array( |
| | 228 | + 'map_group' => 'exif', |
| | 229 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 230 | + 'validate' => 'validateInteger', |
| | 231 | + ), |
| | 232 | + 'Make' => array( |
| | 233 | + 'map_group' => 'exif', |
| | 234 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 235 | + ), |
| | 236 | + 'Model' => array( |
| | 237 | + 'map_group' => 'exif', |
| | 238 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 239 | + ), |
| | 240 | + 'Orientation' => array( |
| | 241 | + 'map_group' => 'exif', |
| | 242 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 243 | + 'validate' => 'validateClosed', |
| | 244 | + 'choices' => array( '1' => true, '2' => true, '3' => true, '4' => true, 5 => true, |
| | 245 | + '6' => true, '7' => true, '8' => true ), |
| | 246 | + ), |
| | 247 | + 'PhotometricInterpretation' => array( |
| | 248 | + 'map_group' => 'exif', |
| | 249 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 250 | + 'validate' => 'validateClosed', |
| | 251 | + 'choices' => array( '2' => true, '6' => true ), |
| | 252 | + ), |
| | 253 | + 'PlanerConfiguration' => array( |
| | 254 | + 'map_group' => 'exif', |
| | 255 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 256 | + 'validate' => 'validateClosed', |
| | 257 | + 'choices' => array( '1' => true, '2' => true ), |
| | 258 | + ), |
| | 259 | + 'PrimaryChromaticities' => array( |
| | 260 | + 'map_group' => 'exif', |
| | 261 | + 'mode' => XMPReader::MODE_SEQ, |
| | 262 | + 'validate' => 'validateRational', |
| | 263 | + ), |
| | 264 | + 'ReferenceBlackWhite' => array( |
| | 265 | + 'map_group' => 'exif', |
| | 266 | + 'mode' => XMPReader::MODE_SEQ, |
| | 267 | + 'validate' => 'validateRational', |
| | 268 | + ), |
| | 269 | + 'ResolutionUnit' => array( |
| | 270 | + 'map_group' => 'exif', |
| | 271 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 272 | + 'validate' => 'validateClosed', |
| | 273 | + 'choices' => array( '2' => true, '3' => true ), |
| | 274 | + ), |
| | 275 | + 'SamplesPerPixel' => array( |
| | 276 | + 'map_group' => 'exif', |
| | 277 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 278 | + 'validate' => 'validateInteger', |
| | 279 | + ), |
| | 280 | + 'Software' => array( /* see xmp:CreatorTool */ |
| | 281 | + 'map_group' => 'exif', |
| | 282 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 283 | + ), |
| | 284 | + /* ignore TransferFunction */ |
| | 285 | + 'WhitePoint' => array( |
| | 286 | + 'map_group' => 'exif', |
| | 287 | + 'mode' => XMPReader::MODE_SEQ, |
| | 288 | + 'validate' => 'validateRational', |
| | 289 | + ), |
| | 290 | + 'XResolution' => array( |
| | 291 | + 'map_group' => 'exif', |
| | 292 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 293 | + 'validate' => 'validateRational', |
| | 294 | + ), |
| | 295 | + 'YResolution' => array( |
| | 296 | + 'map_group' => 'exif', |
| | 297 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 298 | + 'validate' => 'validateRational', |
| | 299 | + ), |
| | 300 | + 'YCbCrCoefficients' => array( |
| | 301 | + 'map_group' => 'exif', |
| | 302 | + 'mode' => XMPReader::MODE_SEQ, |
| | 303 | + 'validate' => 'validateRational', |
| | 304 | + ), |
| | 305 | + 'YCbCrPositioning' => array( |
| | 306 | + 'map_group' => 'exif', |
| | 307 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 308 | + 'validate' => 'validateClosed', |
| | 309 | + 'choices' => array( '1' => true, '2' => true ), |
| | 310 | + ), |
| | 311 | + 'YCbCrSubSampling' => array( |
| | 312 | + 'map_group' => 'exif', |
| | 313 | + 'mode' => XMPReader::MODE_SEQ, |
| | 314 | + 'validate' => 'validateClosed', |
| | 315 | + 'choices' => array( '1' => true, '2' => true ), |
| | 316 | + ), |
| | 317 | + ), |
| | 318 | + 'http://ns.adobe.com/exif/1.0/aux/' => array( |
| | 319 | + 'Lens' => array( |
| | 320 | + 'map_group' => 'exif', |
| | 321 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 322 | + ), |
| | 323 | + 'SerialNumber' => array( |
| | 324 | + 'map_group' => 'exif', |
| | 325 | + 'mode' => XMPReader::MODE_SIMPLE, |
| | 326 | + ), |
| | 327 | + ), |
| 192 | 328 | 'http://purl.org/dc/elements/1.1/' => array( |
| 193 | 329 | 'title' => array( |
| 194 | 330 | 'map_group' => 'general', |
| Index: branches/img_metadata/phase3/includes/media/PNGMetadataExtractor.php |
| — | — | @@ -17,6 +17,8 @@ |
| 18 | 18 | $frameCount = 0; |
| 19 | 19 | $loopCount = 1; |
| 20 | 20 | $duration = 0.0; |
| | 21 | + $xmp = ''; |
| | 22 | + $meta = array(); |
| 21 | 23 | |
| 22 | 24 | if (!$filename) |
| 23 | 25 | throw new Exception( __METHOD__ . ": No file name specified" ); |
| — | — | @@ -61,9 +63,23 @@ |
| 62 | 64 | if( $fctldur['delay_num'] ) { |
| 63 | 65 | $duration += $fctldur['delay_num'] / $fctldur['delay_den']; |
| 64 | 66 | } |
| 65 | | - } elseif ( ( $chunk_type == "IDAT" || $chunk_type == "IEND" ) && $frameCount == 0 ) { |
| 66 | | - // Not a valid animated image. No point in continuing. |
| 67 | | - break; |
| | 67 | + } elseif ( $chunk_type == "iTXt" ) { |
| | 68 | + // At the moment this only does XMP iText chunks, |
| | 69 | + // but in the future might extract other metadata chunks. |
| | 70 | + if( $chunk_size <= 22 ) { |
| | 71 | + // something weird, so skip |
| | 72 | + fseek( $fh, $chunk_size, SEEK_CUR ); |
| | 73 | + continue; |
| | 74 | + } |
| | 75 | + $itxtHeader = fread( $fh, 22 ); |
| | 76 | + if( !$itxtHeader ) { throw new Exception( __METHOD__ . ": Read error" ); return; } |
| | 77 | + if( $itxtHeader !== "XML:com.adobe.xmp\x00\x00\x00\x00\x00" ) { |
| | 78 | + // some other iTXt chunk. |
| | 79 | + fseek( $fh, $chunk_size - 22, SEEK_CUR ); |
| | 80 | + continue; |
| | 81 | + } |
| | 82 | + $xmp = fread( $fh, $chunk_size - 22 ); |
| | 83 | + if( !$xmp ) { throw new Exception( __METHOD__ . ": Read error" ); return; } |
| 68 | 84 | } elseif ( $chunk_type == "IEND" ) { |
| 69 | 85 | break; |
| 70 | 86 | } else { |
| — | — | @@ -80,7 +96,8 @@ |
| 81 | 97 | return array( |
| 82 | 98 | 'frameCount' => $frameCount, |
| 83 | 99 | 'loopCount' => $loopCount, |
| 84 | | - 'duration' => $duration |
| | 100 | + 'duration' => $duration, |
| | 101 | + 'xmp' => $xmp, |
| 85 | 102 | ); |
| 86 | 103 | |
| 87 | 104 | } |
| Index: branches/img_metadata/phase3/includes/media/JpegMetadataExtractor.php |
| — | — | @@ -0,0 +1,205 @@ |
| | 2 | +<?php |
| | 3 | +/** |
| | 4 | +* Class for reading jpegs and extracting metadata. |
| | 5 | +* see also BitmapMetadataHandler. |
| | 6 | +* |
| | 7 | +* Based somewhat on GIFMetadataExtrator. |
| | 8 | +*/ |
| | 9 | +class JpegMetadataExtractor { |
| | 10 | + const MAX_JPEG_SEGMENTS = 200; |
| | 11 | + // the max segment is a sanity check. |
| | 12 | + // A jpeg file should never even remotely have |
| | 13 | + // that many segments. Your average file has about 10. |
| | 14 | + |
| | 15 | + /** Function to extract metadata segmants of interest from jpeg files |
| | 16 | + * based on GIFMetadataExtractor. |
| | 17 | + * |
| | 18 | + * we can almost use getimagesize to do this |
| | 19 | + * but gis doesn't support having multiple app1 segments |
| | 20 | + * and those can't extract xmp on files containing both exif and xmp data |
| | 21 | + * |
| | 22 | + * @param String $filename name of jpeg file |
| | 23 | + * @return Array of interesting segments. |
| | 24 | + * @throws MWException if given invalid file. |
| | 25 | + */ |
| | 26 | + static function segmentSplitter ($filename) { |
| | 27 | + global $wgShowXMP; |
| | 28 | + |
| | 29 | + $segmentCount = 0; |
| | 30 | + |
| | 31 | + $segments = Array( 'XMP_ext' => array(), 'COM' => array() ); |
| | 32 | + |
| | 33 | + if ( !$filename ) throw new MWException( "No filename specified for " . __METHOD__ ); |
| | 34 | + if ( !file_exists( $filename ) || is_dir( $filename ) ) throw new MWException( "Invalid file $filename passed to " . __METHOD__ ); |
| | 35 | + |
| | 36 | + $fh = fopen( $filename, "rb" ); |
| | 37 | + |
| | 38 | + if ( !$fh ) throw new MWException( "Could not open file $filename" ); |
| | 39 | + |
| | 40 | + $buffer = fread( $fh, 2 ); |
| | 41 | + if ( $buffer !== "\xFF\xD8" ) throw new MWException( "Not a jpeg, no SOI" ); |
| | 42 | + while ( !feof( $fh ) ) { |
| | 43 | + $buffer = fread( $fh, 1 ); |
| | 44 | + $segmentCount++; |
| | 45 | + if ( $segmentCount > self::MAX_JPEG_SEGMENTS ) { |
| | 46 | + // this is just a sanity check |
| | 47 | + throw new MWException('Too many jpeg segments. Aborting'); |
| | 48 | + } |
| | 49 | + if ( $buffer !== "\xFF" ) { |
| | 50 | + throw new MWException( "Error reading jpeg file marker" ); |
| | 51 | + } |
| | 52 | + |
| | 53 | + $buffer = fread( $fh, 1 ); |
| | 54 | + if ( $buffer === "\xFE" ) { |
| | 55 | + |
| | 56 | + // COM section -- file comment |
| | 57 | + // First see if valid utf-8, |
| | 58 | + // if not try to convert it to windows-1252. |
| | 59 | + $com = $oldCom = trim( self::jpegExtractMarker( $fh ) ); |
| | 60 | + |
| | 61 | + UtfNormal::quickIsNFCVerify( $com ); |
| | 62 | + //turns $com to valid utf-8. |
| | 63 | + //thus if no change, its utf-8, otherwise its something else. |
| | 64 | + if ( $com !== $oldCom ) { |
| | 65 | + $oldCom = iconv( 'windows-1252', 'UTF-8//IGNORE', $oldCom ); |
| | 66 | + } |
| | 67 | + $segments["COM"][] = $oldCom; |
| | 68 | + |
| | 69 | + } elseif ( $buffer === "\xE1" && $wgShowXMP ) { |
| | 70 | + // APP1 section (Exif, XMP, and XMP extended) |
| | 71 | + // only extract if XMP is enabled. |
| | 72 | + $temp = self::jpegExtractMarker( $fh ); |
| | 73 | + |
| | 74 | + // check what type of app segment this is. |
| | 75 | + if ( substr( $temp, 0, 29 ) === "http://ns.adobe.com/xap/1.0/\x00" ) { |
| | 76 | + $segments["XMP"] = substr( $temp, 29 ); |
| | 77 | + } elseif ( substr( $temp, 0, 35 ) === "http://ns.adobe.com/xmp/extension/\x00" ) { |
| | 78 | + $segments["XMP_ext"][] = substr( $temp, 35 ); |
| | 79 | + } |
| | 80 | + } elseif ( $buffer === "\xED" ) { |
| | 81 | + // APP13 - PSIR. IPTC and some photoshop stuff |
| | 82 | + $temp = self::jpegExtractMarker( $fh ); |
| | 83 | + if ( substr( $temp, 0, 14 ) === "Photoshop 3.0\x00" ) { |
| | 84 | + $segments["PSIR"] = $temp; |
| | 85 | + } |
| | 86 | + } elseif ( $buffer === "\xD9" || $buffer === "\xDA" ) { |
| | 87 | + // EOI - end of image or SOS - start of scan. either way we're past any interesting segments |
| | 88 | + return $segments; |
| | 89 | + } else { |
| | 90 | + // segment we don't care about, so skip |
| | 91 | + $size = unpack( "nint", fread( $fh, 2 ) ); |
| | 92 | + if ( $size['int'] <= 2 ) throw new MWException( "invalid marker size in jpeg" ); |
| | 93 | + fseek( $fh, $size['int'] - 2, SEEK_CUR ); |
| | 94 | + } |
| | 95 | + |
| | 96 | + } |
| | 97 | + // shouldn't get here. |
| | 98 | + throw new MWException( "Reached end of jpeg file unexpectedly" ); |
| | 99 | + |
| | 100 | + } |
| | 101 | + /** |
| | 102 | + * Helper function for jpegSegmentSplitter |
| | 103 | + * @param &$fh FileHandle for jpeg file |
| | 104 | + * @return data content of segment. |
| | 105 | + */ |
| | 106 | + private static function jpegExtractMarker( &$fh ) { |
| | 107 | + $size = unpack( "nint", fread( $fh, 2 ) ); |
| | 108 | + if ( $size['int'] <= 2 ) throw new MWException( "invalid marker size in jpeg" ); |
| | 109 | + return fread( $fh, $size['int'] - 2 ); |
| | 110 | + } |
| | 111 | + |
| | 112 | + /** |
| | 113 | + * This reads the photoshop image resource. |
| | 114 | + * Currently it only compares the iptc/iim hash |
| | 115 | + * with the stored hash, which is used to determine the precedence |
| | 116 | + * of the iptc data. In future it may extract some other info, like |
| | 117 | + * url of copyright license. |
| | 118 | + * |
| | 119 | + * This should generally be called by BitmapMetadataHandler::doApp13() |
| | 120 | + * |
| | 121 | + * @param String $app13 photoshop psir app13 block from jpg. |
| | 122 | + * @return String if the iptc hash is good or not. |
| | 123 | + */ |
| | 124 | + public static function doPSIR ( $app13 ) { |
| | 125 | + if ( !$app13 ) return; |
| | 126 | + // First compare hash with real thing |
| | 127 | + // 0x404 contains IPTC, 0x425 has hash |
| | 128 | + // This is used to determine if the iptc is newer than |
| | 129 | + // the xmp data, as xmp programs update the hash, |
| | 130 | + // where non-xmp programs don't. |
| | 131 | + |
| | 132 | + $offset = 14; // skip past PHOTOSHOP 3.0 identifier. should already be checked. |
| | 133 | + $appLen = strlen( $app13 ); |
| | 134 | + $realHash = ""; |
| | 135 | + $recordedHash = ""; |
| | 136 | + |
| | 137 | + // the +12 is the length of an empty item. |
| | 138 | + while ( $offset + 12 <= $appLen ) { |
| | 139 | + $valid = true; |
| | 140 | + $id = false; |
| | 141 | + $lenName = false; |
| | 142 | + $lenData = false; |
| | 143 | + |
| | 144 | + if ( substr( $app13, $offset, 4 ) !== '8BIM' ) { |
| | 145 | + // its supposed to be 8BIM |
| | 146 | + // but apperently sometimes isn't esp. in |
| | 147 | + // really old jpg's |
| | 148 | + $valid = false; |
| | 149 | + } |
| | 150 | + $offset += 4; |
| | 151 | + $id = substr( $app13, $offset, 2 ); |
| | 152 | + // id is a 2 byte id number which identifies |
| | 153 | + // the piece of info this record contains. |
| | 154 | + |
| | 155 | + $offset += 2; |
| | 156 | + |
| | 157 | + // some record types can contain a name, which |
| | 158 | + // is a pascal string 0-padded to be an even |
| | 159 | + // number of bytes. Most times (and any time |
| | 160 | + // we care) this is empty, making it two null bytes. |
| | 161 | + |
| | 162 | + $lenName = ord( substr( $app13, $offset, 1 ) ) + 1; |
| | 163 | + // we never use the name so skip it. +1 for length byte |
| | 164 | + if ( $lenName % 2 == 1 ) $lenName++; // pad to even. |
| | 165 | + $offset += $lenName; |
| | 166 | + |
| | 167 | + // now length of data (unsigned long big endian) |
| | 168 | + $lenData = unpack( 'Nlen', substr( $app13, $offset, 4 ) ); |
| | 169 | + $offset += 4; // 4bytes length field; |
| | 170 | + |
| | 171 | + // this should not happen, but check. |
| | 172 | + if ( $lenData['len'] + $offset > $appLen ) { |
| | 173 | + wfDebug( __METHOD__ . ' PSIR data too long.' ); |
| | 174 | + return 'iptc-no-hash'; |
| | 175 | + } |
| | 176 | + |
| | 177 | + if ( $valid ) { |
| | 178 | + switch ( $id ) { |
| | 179 | + case "\x04\x04": |
| | 180 | + // IPTC block |
| | 181 | + $realHash = md5( substr( $app13, $offset, $lenData['len'] ), true ); |
| | 182 | + break; |
| | 183 | + case "\x04\x25": |
| | 184 | + $recordedHash = substr( $app13, $offset, $lenData['len'] ); |
| | 185 | + break; |
| | 186 | + } |
| | 187 | + } |
| | 188 | + |
| | 189 | + // if odd, add 1 to length to account for |
| | 190 | + // null pad byte. |
| | 191 | + if ( $lenData['len'] % 2 == 1 ) $lenData['len']++; |
| | 192 | + $offset += $lenData['len']; |
| | 193 | + |
| | 194 | + } |
| | 195 | + |
| | 196 | + if ( !$realHash || !$recordedHash ) { |
| | 197 | + return 'iptc-no-hash'; |
| | 198 | + } elseif ( $realHash === $recordedHash ) { |
| | 199 | + return 'iptc-good-hash'; |
| | 200 | + } else { /*$realHash !== $recordedHash */ |
| | 201 | + return 'iptc-bad-hash'; |
| | 202 | + } |
| | 203 | + |
| | 204 | + } |
| | 205 | + |
| | 206 | +} |
| Property changes on: branches/img_metadata/phase3/includes/media/JpegMetadataExtractor.php |
| ___________________________________________________________________ |
| Added: svn:eol-style |
| 1 | 207 | + native |
| Index: branches/img_metadata/phase3/includes/media/PNG.php |
| — | — | @@ -12,22 +12,32 @@ |
| 13 | 13 | class PNGHandler extends BitmapHandler { |
| 14 | 14 | |
| 15 | 15 | function getMetadata( $image, $filename ) { |
| 16 | | - if ( !isset($image->parsedPNGMetadata) ) { |
| 17 | | - try { |
| 18 | | - $image->parsedPNGMetadata = PNGMetadataExtractor::getMetadata( $filename ); |
| 19 | | - } catch( Exception $e ) { |
| 20 | | - // Broken file? |
| 21 | | - wfDebug( __METHOD__ . ': ' . $e->getMessage() . "\n" ); |
| 22 | | - return '0'; |
| 23 | | - } |
| | 16 | + try { |
| | 17 | + $metadata = BitmapMetadataHandler::PNG( $filename ); |
| | 18 | + } catch( Exception $e ) { |
| | 19 | + // Broken file? |
| | 20 | + wfDebug( __METHOD__ . ': ' . $e->getMessage() . "\n" ); |
| | 21 | + return '0'; |
| 24 | 22 | } |
| 25 | 23 | |
| 26 | | - return serialize($image->parsedPNGMetadata); |
| 27 | | - |
| | 24 | + return serialize($metadata); |
| 28 | 25 | } |
| 29 | 26 | |
| 30 | 27 | function formatMetadata( $image ) { |
| 31 | | - return false; |
| | 28 | + $meta = $image->getMetadata(); |
| | 29 | + |
| | 30 | + if ( !$meta ) { |
| | 31 | + return false; |
| | 32 | + } |
| | 33 | + $meta = unserialize( $meta ); |
| | 34 | + if ( !isset( $meta['metadata'] ) || count( $meta['metadata'] ) <= 1 ) { |
| | 35 | + return false; |
| | 36 | + } |
| | 37 | + |
| | 38 | + if ( isset( $meta['metadata']['_MW_PNG_VERSION'] ) ) { |
| | 39 | + unset( $meta['metadata']['_MW_PNG_VERSION'] ); |
| | 40 | + } |
| | 41 | + return $this->formatMetadataHelper( $meta['metadata'] ); |
| 32 | 42 | } |
| 33 | 43 | |
| 34 | 44 | function isAnimatedImage( $image ) { |
| — | — | @@ -47,7 +57,20 @@ |
| 48 | 58 | wfSuppressWarnings(); |
| 49 | 59 | $data = unserialize( $metadata ); |
| 50 | 60 | wfRestoreWarnings(); |
| 51 | | - return (boolean) $data; |
| | 61 | + if ( $data === '0' ) { |
| | 62 | + // Do not repetitivly regenerate metadata on broken file. |
| | 63 | + return self::METADATA_GOOD; |
| | 64 | + } |
| | 65 | + if ( !$data || !is_array( $data ) ) { |
| | 66 | + wfDebug(__METHOD__ . ' invalid png metadata' ); |
| | 67 | + return self::METADATA_BAD; |
| | 68 | + } |
| | 69 | + |
| | 70 | + if ( !isset( $data['metadata']['_MW_PNG_VERSION'] ) ) { |
| | 71 | + wfDebug(__METHOD__ . ' old but compatible png metadata' ); |
| | 72 | + return self::METADATA_COMPATIBLE; |
| | 73 | + } |
| | 74 | + return self::METADATA_GOOD; |
| 52 | 75 | } |
| 53 | 76 | function getLongDesc( $image ) { |
| 54 | 77 | global $wgUser, $wgLang; |
| Index: branches/img_metadata/phase3/includes/media/Jpeg.php |
| — | — | @@ -12,12 +12,13 @@ |
| 13 | 13 | |
| 14 | 14 | function getMetadata ( $image, $filename ) { |
| 15 | 15 | try { |
| 16 | | - $meta = BitmapMetadataHandler::newForJpeg( $filename ); |
| 17 | | - $temp = $meta->getMetadataArray(); |
| 18 | | - if ( $temp ) { |
| 19 | | - $temp['MEDIAWIKI_EXIF_VERSION'] = Exif::version(); |
| 20 | | - return serialize( $temp ); |
| | 16 | + $meta = BitmapMetadataHandler::Jpeg( $filename ); |
| | 17 | + if ( $meta ) { |
| | 18 | + $meta['MEDIAWIKI_EXIF_VERSION'] = Exif::version(); |
| | 19 | + return serialize( $meta ); |
| 21 | 20 | } else { |
| | 21 | + /* FIXME, this should probably be something else to do versioning |
| | 22 | + with older files that say have no exif, but have xmp */ |
| 22 | 23 | return '0'; |
| 23 | 24 | } |
| 24 | 25 | } |
| Index: branches/img_metadata/phase3/includes/AutoLoader.php |
| — | — | @@ -455,6 +455,7 @@ |
| 456 | 456 | 'PNGMetadataExtractor' => 'includes/media/PNGMetadataExtractor.php', |
| 457 | 457 | 'SvgHandler' => 'includes/media/SVG.php', |
| 458 | 458 | 'JpegHandler' => 'includes/media/Jpeg.php', |
| | 459 | + 'JpegMetadataExtractor' => 'includes/media/JpegMetadataExtractor.php', |
| 459 | 460 | 'BitmapMetadataHandler' => 'includes/media/BitmapMetadataHandler.php', |
| 460 | 461 | 'IPTC' => 'includes/media/IPTC.php', |
| 461 | 462 | 'ThumbnailImage' => 'includes/media/MediaTransformOutput.php', |
| Index: branches/img_metadata/phase3/includes/DefaultSettings.php |
| — | — | @@ -388,6 +388,11 @@ |
| 389 | 389 | $wgShowEXIF = function_exists( 'exif_read_data' ); |
| 390 | 390 | |
| 391 | 391 | /** |
| | 392 | + * Show/extract XMP metadata (similar to above exif setting) |
| | 393 | + */ |
| | 394 | +$wgShowXMP = function_exists( 'xml_parser_create_ns' ); |
| | 395 | + |
| | 396 | +/** |
| 392 | 397 | * If to automatically update the img_metadata field |
| 393 | 398 | * if the metadata field is outdated but compatible with the current version. |
| 394 | 399 | * Defaults to false. |
| Index: branches/img_metadata/phase3/languages/messages/MessagesEn.php |
| — | — | @@ -3674,7 +3674,6 @@ |
| 3675 | 3675 | 'exif-stripbytecounts' => 'Bytes per compressed strip', |
| 3676 | 3676 | 'exif-jpeginterchangeformat' => 'Offset to JPEG SOI', |
| 3677 | 3677 | 'exif-jpeginterchangeformatlength' => 'Bytes of JPEG data', |
| 3678 | | -'exif-transferfunction' => 'Transfer function', |
| 3679 | 3678 | 'exif-whitepoint' => 'White point chromaticity', |
| 3680 | 3679 | 'exif-primarychromaticities' => 'Chromaticities of primarities', |
| 3681 | 3680 | 'exif-ycbcrcoefficients' => 'Color space transformation matrix coefficients', |
| — | — | @@ -3804,6 +3803,8 @@ |
| 3805 | 3804 | 'exif-datetimeexpires' => 'Do not use after', |
| 3806 | 3805 | 'exif-datetimereleased' => 'Released on', |
| 3807 | 3806 | 'exif-originaltransmissionref' => 'Original transmission location code', |
| | 3807 | +'exif-lens' => 'Lens used', |
| | 3808 | +'exif-serialnumber' => 'Serial number of camera', |
| 3808 | 3809 | |
| 3809 | 3810 | # Make & model, can be wikified in order to link to the camera and model name |
| 3810 | 3811 | 'exif-make-value' => '$1', # do not translate or duplicate this message to other languages |