diff -Naur org/apache/tika/parser/image/ImageMetadataExtractor.java org/apache/tika/parser/image/ImageMetadataExtractor.java --- org/apache/tika/parser/image/ImageMetadataExtractor.java 2012-07-10 12:01:58.000000000 +0200 +++ org/apache/tika/parser/image/ImageMetadataExtractor.java 2012-09-21 14:43:57.346650852 +0200 @@ -37,6 +37,7 @@ import org.apache.tika.metadata.TikaCoreProperties; import org.xml.sax.SAXException; +import com.drew.imaging.jpeg.JpegMetadataReader; import com.drew.imaging.jpeg.JpegProcessingException; import com.drew.imaging.jpeg.JpegSegmentReader; import com.drew.imaging.tiff.TiffMetadataReader; @@ -46,8 +47,10 @@ import com.drew.metadata.MetadataException; import com.drew.metadata.MetadataReader; import com.drew.metadata.Tag; -import com.drew.metadata.exif.ExifDirectory; +import com.drew.metadata.exif.ExifIFD0Directory; import com.drew.metadata.exif.ExifReader; +import com.drew.metadata.exif.ExifSubIFDDirectory; +import com.drew.metadata.exif.ExifThumbnailDirectory; import com.drew.metadata.exif.GpsDirectory; import com.drew.metadata.iptc.IptcDirectory; import com.drew.metadata.iptc.IptcReader; @@ -111,19 +114,12 @@ JpegSegmentReader reader, byte marker, Class klass) { try { - Constructor constructor = - klass.getConstructor(byte[].class); - int n = reader.getSegmentCount(marker); - for (int i = 0; i < n; i++) { - byte[] segment = reader.readSegment(marker, i); - - com.drew.metadata.Metadata metadata = - new com.drew.metadata.Metadata(); - constructor.newInstance(segment).extract(metadata); - handle(metadata); - } + com.drew.metadata.Metadata metadata = JpegMetadataReader.extractMetadataFromJpegSegmentReader(reader.getSegmentData()); + new com.drew.metadata.Metadata(); + + handle(metadata); } catch (Exception e) { // Unable to read this kind of metadata, so skip } @@ -133,11 +129,9 @@ throws IOException, SAXException, TikaException { try { com.drew.metadata.Metadata tiffMetadata = - TiffMetadataReader.readMetadata(stream); + TiffMetadataReader.readMetadata(stream, false); handle(tiffMetadata); - } catch (TiffProcessingException e) { - throw new TikaException("Can't read TIFF metadata", e); } catch (MetadataException e) { throw new TikaException("Can't read TIFF metadata", e); } @@ -151,7 +145,7 @@ @SuppressWarnings("unchecked") protected void handle(com.drew.metadata.Metadata metadataExtractor) throws MetadataException { - handle(metadataExtractor.getDirectoryIterator()); + handle(metadataExtractor.getDirectories().iterator()); } /** @@ -198,7 +192,7 @@ } public void handle(Directory directory, Metadata metadata) throws MetadataException { - Iterator tags = directory.getTagIterator(); + Iterator tags = directory.getTags().iterator(); while (tags.hasNext()) { Tag tag = (Tag) tags.next(); metadata.set(tag.getTagName(), tag.getDescription()); @@ -217,23 +211,18 @@ } public void handle(Directory directory, Metadata metadata) throws MetadataException { - Iterator tags = directory.getTagIterator(); + Iterator tags = directory.getTags().iterator(); while (tags.hasNext()) { Tag tag = (Tag) tags.next(); String name = tag.getTagName(); if (!MetadataFields.isMetadataField(name)) { - try { - String value = tag.getDescription().trim(); - if (Boolean.TRUE.toString().equalsIgnoreCase(value)) { - value = Boolean.TRUE.toString(); - } else if (Boolean.FALSE.toString().equalsIgnoreCase(value)) { - value = Boolean.FALSE.toString(); - } - metadata.set(name, value); - } catch(MetadataException e) { - // Either something's corrupt, or it's a JPEG tag - // that the library doesn't know about. Skip it + String value = tag.getDescription().trim(); + if (Boolean.TRUE.toString().equalsIgnoreCase(value)) { + value = Boolean.TRUE.toString(); + } else if (Boolean.FALSE.toString().equalsIgnoreCase(value)) { + value = Boolean.FALSE.toString(); } + metadata.set(name, value); } } } @@ -245,22 +234,22 @@ static class DimensionsHandler implements DirectoryHandler { private final Pattern LEADING_NUMBERS = Pattern.compile("(\\d+)\\s*.*"); public boolean supports(Class directoryType) { - return directoryType == JpegDirectory.class || directoryType == ExifDirectory.class; + return directoryType == JpegDirectory.class || directoryType == ExifThumbnailDirectory.class; } public void handle(Directory directory, Metadata metadata) throws MetadataException { // The test TIFF has width and height stored as follows according to exiv2 //Exif.Image.ImageWidth Short 1 100 //Exif.Image.ImageLength Short 1 75 // and the values are found in "Thumbnail Image Width" (and Height) from Metadata Extractor - set(directory, metadata, ExifDirectory.TAG_THUMBNAIL_IMAGE_WIDTH, Metadata.IMAGE_WIDTH); + set(directory, metadata, ExifThumbnailDirectory.TAG_THUMBNAIL_IMAGE_WIDTH, Metadata.IMAGE_WIDTH); set(directory, metadata, JpegDirectory.TAG_JPEG_IMAGE_WIDTH, Metadata.IMAGE_WIDTH); - set(directory, metadata, ExifDirectory.TAG_THUMBNAIL_IMAGE_HEIGHT, Metadata.IMAGE_LENGTH); + set(directory, metadata, ExifThumbnailDirectory.TAG_THUMBNAIL_IMAGE_HEIGHT, Metadata.IMAGE_LENGTH); set(directory, metadata, JpegDirectory.TAG_JPEG_IMAGE_HEIGHT, Metadata.IMAGE_LENGTH); // Bits per sample, two methods of extracting, exif overrides jpeg set(directory, metadata, JpegDirectory.TAG_JPEG_DATA_PRECISION, Metadata.BITS_PER_SAMPLE); - set(directory, metadata, ExifDirectory.TAG_BITS_PER_SAMPLE, Metadata.BITS_PER_SAMPLE); + set(directory, metadata, ExifThumbnailDirectory.TAG_BITS_PER_SAMPLE, Metadata.BITS_PER_SAMPLE); // Straightforward - set(directory, metadata, ExifDirectory.TAG_SAMPLES_PER_PIXEL, Metadata.SAMPLES_PER_PIXEL); + set(directory, metadata, ExifThumbnailDirectory.TAG_SAMPLES_PER_PIXEL, Metadata.SAMPLES_PER_PIXEL); } private void set(Directory directory, Metadata metadata, int extractTag, Property metadataField) { if (directory.containsTag(extractTag)) { @@ -286,7 +275,7 @@ static class ExifHandler implements DirectoryHandler { private static final SimpleDateFormat DATE_UNSPECIFIED_TZ = new SimpleDateFormat("yyyy-MM-dd'T'HH:mm:ss"); public boolean supports(Class directoryType) { - return directoryType == ExifDirectory.class; + return directoryType == ExifIFD0Directory.class || directoryType == ExifSubIFDDirectory.class; } public void handle(Directory directory, Metadata metadata) { try { @@ -303,8 +292,8 @@ */ public void handleCommentTags(Directory directory, Metadata metadata) { if (metadata.get(TikaCoreProperties.DESCRIPTION) == null && - directory.containsTag(ExifDirectory.TAG_IMAGE_DESCRIPTION)) { - metadata.set(TikaCoreProperties.DESCRIPTION, directory.getString(ExifDirectory.TAG_IMAGE_DESCRIPTION)); + directory.containsTag(ExifIFD0Directory.TAG_IMAGE_DESCRIPTION)) { + metadata.set(TikaCoreProperties.DESCRIPTION, directory.getString(ExifIFD0Directory.TAG_IMAGE_DESCRIPTION)); } } /** @@ -312,22 +301,18 @@ * TIFF image metadata namespace. */ public void handlePhotoTags(Directory directory, Metadata metadata) { - if(directory.containsTag(ExifDirectory.TAG_EXPOSURE_TIME)) { - Object exposure = directory.getObject(ExifDirectory.TAG_EXPOSURE_TIME); + if(directory.containsTag(ExifSubIFDDirectory.TAG_EXPOSURE_TIME)) { + Object exposure = directory.getObject(ExifSubIFDDirectory.TAG_EXPOSURE_TIME); if(exposure instanceof Rational) { metadata.set(Metadata.EXPOSURE_TIME, ((Rational)exposure).doubleValue()); } else { - metadata.set(Metadata.EXPOSURE_TIME, directory.getString(ExifDirectory.TAG_EXPOSURE_TIME)); + metadata.set(Metadata.EXPOSURE_TIME, directory.getString(ExifSubIFDDirectory.TAG_EXPOSURE_TIME)); } } - if(directory.containsTag(ExifDirectory.TAG_FLASH)) { + if(directory.containsTag(ExifSubIFDDirectory.TAG_FLASH)) { String flash = ""; - try { - flash = directory.getDescription(ExifDirectory.TAG_FLASH); - } catch (MetadataException e) { - // ignore - } + flash = directory.getDescription(ExifSubIFDDirectory.TAG_FLASH); if(flash.indexOf("Flash fired") > -1) { metadata.set(Metadata.FLASH_FIRED, Boolean.TRUE.toString()); } @@ -339,70 +324,66 @@ } } - if(directory.containsTag(ExifDirectory.TAG_FNUMBER)) { - Object fnumber = directory.getObject(ExifDirectory.TAG_FNUMBER); + if(directory.containsTag(ExifSubIFDDirectory.TAG_FNUMBER)) { + Object fnumber = directory.getObject(ExifSubIFDDirectory.TAG_FNUMBER); if(fnumber instanceof Rational) { metadata.set(Metadata.F_NUMBER, ((Rational)fnumber).doubleValue()); } else { - metadata.set(Metadata.F_NUMBER, directory.getString(ExifDirectory.TAG_FNUMBER)); + metadata.set(Metadata.F_NUMBER, directory.getString(ExifSubIFDDirectory.TAG_FNUMBER)); } } - if(directory.containsTag(ExifDirectory.TAG_FOCAL_LENGTH)) { - Object length = directory.getObject(ExifDirectory.TAG_FOCAL_LENGTH); + if(directory.containsTag(ExifSubIFDDirectory.TAG_FOCAL_LENGTH)) { + Object length = directory.getObject(ExifSubIFDDirectory.TAG_FOCAL_LENGTH); if(length instanceof Rational) { metadata.set(Metadata.FOCAL_LENGTH, ((Rational)length).doubleValue()); } else { - metadata.set(Metadata.FOCAL_LENGTH, directory.getString(ExifDirectory.TAG_FOCAL_LENGTH)); + metadata.set(Metadata.FOCAL_LENGTH, directory.getString(ExifSubIFDDirectory.TAG_FOCAL_LENGTH)); } } - if(directory.containsTag(ExifDirectory.TAG_ISO_EQUIVALENT)) { - metadata.set(Metadata.ISO_SPEED_RATINGS, directory.getString(ExifDirectory.TAG_ISO_EQUIVALENT)); + if(directory.containsTag(ExifSubIFDDirectory.TAG_ISO_EQUIVALENT)) { + metadata.set(Metadata.ISO_SPEED_RATINGS, directory.getString(ExifSubIFDDirectory.TAG_ISO_EQUIVALENT)); } - if(directory.containsTag(ExifDirectory.TAG_MAKE)) { - metadata.set(Metadata.EQUIPMENT_MAKE, directory.getString(ExifDirectory.TAG_MAKE)); + if(directory.containsTag(ExifIFD0Directory.TAG_MAKE)) { + metadata.set(Metadata.EQUIPMENT_MAKE, directory.getString(ExifIFD0Directory.TAG_MAKE)); } - if(directory.containsTag(ExifDirectory.TAG_MODEL)) { - metadata.set(Metadata.EQUIPMENT_MODEL, directory.getString(ExifDirectory.TAG_MODEL)); + if(directory.containsTag(ExifIFD0Directory.TAG_MODEL)) { + metadata.set(Metadata.EQUIPMENT_MODEL, directory.getString(ExifIFD0Directory.TAG_MODEL)); } - if(directory.containsTag(ExifDirectory.TAG_ORIENTATION)) { - Object length = directory.getObject(ExifDirectory.TAG_ORIENTATION); + if(directory.containsTag(ExifIFD0Directory.TAG_ORIENTATION)) { + Object length = directory.getObject(ExifIFD0Directory.TAG_ORIENTATION); if(length instanceof Integer) { metadata.set(Metadata.ORIENTATION, Integer.toString( ((Integer)length).intValue() )); } else { - metadata.set(Metadata.ORIENTATION, directory.getString(ExifDirectory.TAG_ORIENTATION)); + metadata.set(Metadata.ORIENTATION, directory.getString(ExifIFD0Directory.TAG_ORIENTATION)); } } - if(directory.containsTag(ExifDirectory.TAG_SOFTWARE)) { - metadata.set(Metadata.SOFTWARE, directory.getString(ExifDirectory.TAG_SOFTWARE)); + if(directory.containsTag(ExifIFD0Directory.TAG_SOFTWARE)) { + metadata.set(Metadata.SOFTWARE, directory.getString(ExifIFD0Directory.TAG_SOFTWARE)); } - if(directory.containsTag(ExifDirectory.TAG_X_RESOLUTION)) { - Object resolution = directory.getObject(ExifDirectory.TAG_X_RESOLUTION); + if(directory.containsTag(ExifIFD0Directory.TAG_X_RESOLUTION)) { + Object resolution = directory.getObject(ExifIFD0Directory.TAG_X_RESOLUTION); if(resolution instanceof Rational) { metadata.set(Metadata.RESOLUTION_HORIZONTAL, ((Rational)resolution).doubleValue()); } else { - metadata.set(Metadata.RESOLUTION_HORIZONTAL, directory.getString(ExifDirectory.TAG_X_RESOLUTION)); + metadata.set(Metadata.RESOLUTION_HORIZONTAL, directory.getString(ExifIFD0Directory.TAG_X_RESOLUTION)); } } - if(directory.containsTag(ExifDirectory.TAG_Y_RESOLUTION)) { - Object resolution = directory.getObject(ExifDirectory.TAG_Y_RESOLUTION); + if(directory.containsTag(ExifIFD0Directory.TAG_Y_RESOLUTION)) { + Object resolution = directory.getObject(ExifIFD0Directory.TAG_Y_RESOLUTION); if(resolution instanceof Rational) { metadata.set(Metadata.RESOLUTION_VERTICAL, ((Rational)resolution).doubleValue()); } else { - metadata.set(Metadata.RESOLUTION_VERTICAL, directory.getString(ExifDirectory.TAG_Y_RESOLUTION)); + metadata.set(Metadata.RESOLUTION_VERTICAL, directory.getString(ExifIFD0Directory.TAG_Y_RESOLUTION)); } } - if(directory.containsTag(ExifDirectory.TAG_RESOLUTION_UNIT)) { - try { - metadata.set(Metadata.RESOLUTION_UNIT, directory.getDescription(ExifDirectory.TAG_RESOLUTION_UNIT)); - } catch (MetadataException e) { - // ignore - } + if(directory.containsTag(ExifIFD0Directory.TAG_RESOLUTION_UNIT)) { + metadata.set(Metadata.RESOLUTION_UNIT, directory.getDescription(ExifIFD0Directory.TAG_RESOLUTION_UNIT)); } } /** @@ -412,16 +393,16 @@ throws MetadataException { // Date/Time Original overrides value from ExifDirectory.TAG_DATETIME Date original = null; - if (directory.containsTag(ExifDirectory.TAG_DATETIME_ORIGINAL)) { - original = directory.getDate(ExifDirectory.TAG_DATETIME_ORIGINAL); + if (directory.containsTag(ExifSubIFDDirectory.TAG_DATETIME_ORIGINAL)) { + original = directory.getDate(ExifSubIFDDirectory.TAG_DATETIME_ORIGINAL); // Unless we have GPS time we don't know the time zone so date must be set // as ISO 8601 datetime without timezone suffix (no Z or +/-) String datetimeNoTimeZone = DATE_UNSPECIFIED_TZ.format(original); // Same time zone as Metadata Extractor uses metadata.set(TikaCoreProperties.CREATED, datetimeNoTimeZone); metadata.set(Metadata.ORIGINAL_DATE, datetimeNoTimeZone); } - if (directory.containsTag(ExifDirectory.TAG_DATETIME)) { - Date datetime = directory.getDate(ExifDirectory.TAG_DATETIME); + if (directory.containsTag(ExifIFD0Directory.TAG_DATETIME)) { + Date datetime = directory.getDate(ExifIFD0Directory.TAG_DATETIME); String datetimeNoTimeZone = DATE_UNSPECIFIED_TZ.format(datetime); metadata.set(TikaCoreProperties.MODIFIED, datetimeNoTimeZone); // If Date/Time Original does not exist this might be creation date