Index: tika-core/src/main/java/org/apache/tika/metadata/XMPDM.java =================================================================== --- tika-core/src/main/java/org/apache/tika/metadata/XMPDM.java (revision 1613819) +++ tika-core/src/main/java/org/apache/tika/metadata/XMPDM.java (working copy) @@ -60,6 +60,21 @@ Property ARTIST = Property.externalText("xmpDM:artist"); /** + * "The name of the album artist or group for compilation albums." + */ + Property ALBUM_ARTIST = Property.externalText("xmpDM:albumArtist"); + + /** + * "The disc number for part of an album set." + */ + Property DISC_NUMBER = Property.externalInteger("xmpDM:discNumber"); + + /** + * "An album created by various artists." + */ + Property COMPILATION = Property.externalInteger("xmpDM:compilation"); + + /** * "The date and time when the audio was last modified." */ Property AUDIO_MOD_DATE = Property.internalDate("xmpDM:audioModDate"); Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/CompositeTagHandler.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/CompositeTagHandler.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/CompositeTagHandler.java (working copy) @@ -113,4 +113,32 @@ return null; } + public String getAlbumArtist() { + for (ID3Tags tag : tags) { + if (tag.getAlbumArtist() != null) { + return tag.getAlbumArtist(); + } + } + return null; + } + + public String getDisc() { + for (ID3Tags tag : tags) { + if (tag.getDisc() != null) { + return tag.getDisc(); + } + } + return null; + } + + public String getCompilation() { + for (ID3Tags tag : tags) { + if (tag.getCompilation() != null) { + return tag.getCompilation(); + } + } + return null; + } + + } Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3Tags.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3Tags.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3Tags.java (working copy) @@ -178,6 +178,12 @@ String getComposer(); + String getAlbumArtist(); // compilation album group + + String getDisc(); // part of set + + String getCompilation(); // part of compilation + /** * Retrieves the comments, if any. * Files may have more than one comment, but normally only Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v1Handler.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v1Handler.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v1Handler.java (working copy) @@ -40,6 +40,10 @@ private ID3Comment comment; private String genre; private String trackNumber; + // No equavalents for v1 + private String albumArtist; + private String disc; + private String compilation; boolean found = false; @@ -120,6 +124,18 @@ return null; } + public String getAlbumArtist() { + return albumArtist; + } + + public String getDisc() { + return disc; + } + + public String getCompilation() { + return compilation; + } + /** * Returns the identified ISO-8859-1 substring from the given byte buffer. * The return value is the zero-terminated substring retrieved from Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v22Handler.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v22Handler.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v22Handler.java (working copy) @@ -39,6 +39,9 @@ private String composer; private String genre; private String trackNumber; + private String albumArtist; + private String disc; + private String compilation; private List comments = new ArrayList(); public ID3v22Handler(ID3v2Frame frame) @@ -50,6 +53,8 @@ title = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TP1")) { artist = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TP2")) { + albumArtist = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TAL")) { album = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TYE")) { @@ -60,6 +65,10 @@ comments.add( getComment(tag.data, 0, tag.data.length) ); } else if (tag.name.equals("TRK")) { trackNumber = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TPA")) { + disc = getTagString(tag.data, 0, tag.data.length); + // } else if (tag.name.equals("???")) { // No tag for ID3v22 + // compilation = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TCO")) { genre = extractGenre( getTagString(tag.data, 0, tag.data.length) ); } @@ -129,6 +138,18 @@ return trackNumber; } + public String getAlbumArtist() { + return albumArtist; + } + + public String getDisc() { + return disc; + } + + public String getCompilation() { + return compilation; + } + private class RawV22TagIterator extends RawTagIterator { private RawV22TagIterator(ID3v2Frame frame) { frame.super(3, 3, 1, 0); Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v23Handler.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v23Handler.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v23Handler.java (working copy) @@ -39,6 +39,9 @@ private String composer; private String genre; private String trackNumber; + private String albumArtist; + private String disc; + private String compilation; private List comments = new ArrayList(); public ID3v23Handler(ID3v2Frame frame) @@ -50,6 +53,8 @@ title = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TPE1")) { artist = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TPE2")) { + albumArtist = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TALB")) { album = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TYER")) { @@ -60,6 +65,10 @@ comments.add( getComment(tag.data, 0, tag.data.length) ); } else if (tag.name.equals("TRCK")) { trackNumber = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TPOS")) { + disc = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TCMP")) { + compilation = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TCON")) { genre = ID3v22Handler.extractGenre( getTagString(tag.data, 0, tag.data.length) ); } @@ -109,6 +118,18 @@ return trackNumber; } + public String getAlbumArtist() { + return albumArtist; + } + + public String getDisc() { + return disc; + } + + public String getCompilation() { + return compilation; + } + private class RawV23TagIterator extends RawTagIterator { private RawV23TagIterator(ID3v2Frame frame) { frame.super(4, 4, 1, 2); Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v24Handler.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v24Handler.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/ID3v24Handler.java (working copy) @@ -40,6 +40,9 @@ private String composer; private String genre; private String trackNumber; + private String albumArtist; + private String disc; + private String compilation; private List comments = new ArrayList(); public ID3v24Handler(ID3v2Frame frame) @@ -51,6 +54,8 @@ title = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TPE1")) { artist = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TPE2")) { + albumArtist = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TALB")) { album = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TYER")) { @@ -65,6 +70,10 @@ comments.add( getComment(tag.data, 0, tag.data.length) ); } else if (tag.name.equals("TRCK")) { trackNumber = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TPOS")) { + disc = getTagString(tag.data, 0, tag.data.length); + } else if (tag.name.equals("TCMP")) { + compilation = getTagString(tag.data, 0, tag.data.length); } else if (tag.name.equals("TCON")) { genre = ID3v22Handler.extractGenre( getTagString(tag.data, 0, tag.data.length) ); } @@ -114,7 +123,19 @@ return trackNumber; } - private class RawV24TagIterator extends RawTagIterator { + public String getAlbumArtist() { + return albumArtist; + } + + public String getDisc() { + return disc; + } + + public String getCompilation() { + return compilation; + } + + private class RawV24TagIterator extends RawTagIterator { private RawV24TagIterator(ID3v2Frame frame) { frame.super(4, 4, 1, 2); } Index: tika-parsers/src/main/java/org/apache/tika/parser/mp3/Mp3Parser.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp3/Mp3Parser.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp3/Mp3Parser.java (working copy) @@ -81,6 +81,9 @@ metadata.set(XMPDM.RELEASE_DATE, tag.getYear()); metadata.set(XMPDM.GENRE, tag.getGenre()); metadata.set(XMPDM.DURATION, audioAndTags.duration); + metadata.set(XMPDM.ALBUM_ARTIST, tag.getAlbumArtist()); + metadata.set(XMPDM.DISC_NUMBER, tag.getDisc()); + metadata.set(XMPDM.COMPILATION, tag.getCompilation()); List comments = new ArrayList(); for (ID3Comment comment : tag.getComments()) { Index: tika-parsers/src/main/java/org/apache/tika/parser/mp4/MP4Parser.java =================================================================== --- tika-parsers/src/main/java/org/apache/tika/parser/mp4/MP4Parser.java (revision 1613819) +++ tika-parsers/src/main/java/org/apache/tika/parser/mp4/MP4Parser.java (working copy) @@ -55,7 +55,10 @@ import com.coremedia.iso.boxes.sampleentry.AudioSampleEntry; import com.googlecode.mp4parser.boxes.apple.AppleAlbumBox; import com.googlecode.mp4parser.boxes.apple.AppleArtistBox; +import com.googlecode.mp4parser.boxes.apple.AppleArtist2Box; import com.googlecode.mp4parser.boxes.apple.AppleCommentBox; +import com.googlecode.mp4parser.boxes.apple.AppleCompilationBox; +import com.googlecode.mp4parser.boxes.apple.AppleDiskNumberBox; import com.googlecode.mp4parser.boxes.apple.AppleEncoderBox; import com.googlecode.mp4parser.boxes.apple.AppleGenreBox; import com.googlecode.mp4parser.boxes.apple.AppleNameBox; @@ -205,7 +208,7 @@ // Check for iTunes Metadata // See http://atomicparsley.sourceforge.net/mpeg-4files.html and - // http://code.google.com/p/mp4v2/wiki/iTunesMetadata for more on these + // http://code.google.com/p/mp4v2/wiki/iTunesMetadata for more on these AppleItemListBox apple = getOrNull(meta, AppleItemListBox.class); if (apple != null) { // Title @@ -217,6 +220,10 @@ addMetadata(TikaCoreProperties.CREATOR, metadata, artist); addMetadata(XMPDM.ARTIST, metadata, artist); + // Album Artist + AppleArtist2Box artist2 = getOrNull(apple, AppleArtist2Box.class); + addMetadata(XMPDM.ALBUM_ARTIST, metadata, artist2); + // Album AppleAlbumBox album = getOrNull(apple, AppleAlbumBox.class); addMetadata(XMPDM.ALBUM, metadata, album); @@ -242,6 +249,20 @@ //metadata.set(XMPDM.NUMBER_OF_TRACKS, trackNum.getB()); // TODO } + // Disc number + AppleDiskNumberBox discNum = getOrNull(apple, AppleDiskNumberBox.class); + if (discNum != null) { + metadata.set(XMPDM.DISC_NUMBER, discNum.getA()); + } + + // Compilation + AppleCompilationBox compilation = getOrNull(apple, AppleCompilationBox.class); + if (compilation != null) { + Object obj = compilation.getValue(); + System.out.println( "Obj type=" + obj.getClass().getName() + ", value=" +compilation.getValue()); + metadata.set(XMPDM.COMPILATION, (int) compilation.getValue()); + } + // Comment AppleCommentBox comment = getOrNull(apple, AppleCommentBox.class); addMetadata(XMPDM.LOG_COMMENT, metadata, comment); Index: tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java =================================================================== --- tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java (revision 1613819) +++ tika-parsers/src/test/java/org/apache/tika/parser/mp3/Mp3ParserTest.java (working copy) @@ -124,11 +124,14 @@ // Check XMPDM-typed audio properties assertEquals("Test Album", metadata.get(XMPDM.ALBUM)); assertEquals("Test Artist", metadata.get(XMPDM.ARTIST)); + assertEquals("Test Album Artist", metadata.get(XMPDM.ALBUM_ARTIST)); assertEquals(null, metadata.get(XMPDM.COMPOSER)); assertEquals("2008", metadata.get(XMPDM.RELEASE_DATE)); assertEquals("Rock", metadata.get(XMPDM.GENRE)); assertEquals("XXX - ID3v1 Comment\nTest Comment", metadata.get(XMPDM.LOG_COMMENT.getName())); assertEquals("1", metadata.get(XMPDM.TRACK_NUMBER)); + assertEquals("1/1", metadata.get(XMPDM.DISC_NUMBER)); + assertEquals("1", metadata.get(XMPDM.COMPILATION)); assertEquals("44100", metadata.get(XMPDM.AUDIO_SAMPLE_RATE)); assertEquals("Mono", metadata.get(XMPDM.AUDIO_CHANNEL_TYPE)); @@ -208,6 +211,16 @@ assertEquals("44100", metadata.get("samplerate")); assertEquals("1", metadata.get("channels")); checkDuration(metadata, 2); + + // Check XMPDM-typed audio properties + assertEquals("Test Album", metadata.get(XMPDM.ALBUM)); + assertEquals("Test Artist", metadata.get(XMPDM.ARTIST)); + assertEquals("Test Album Artist", metadata.get(XMPDM.ALBUM_ARTIST)); + assertEquals(null, metadata.get(XMPDM.COMPOSER)); + assertEquals("2008", metadata.get(XMPDM.RELEASE_DATE)); + assertEquals("Rock", metadata.get(XMPDM.GENRE)); + assertEquals("1", metadata.get(XMPDM.DISC_NUMBER)); + assertEquals("1", metadata.get(XMPDM.COMPILATION)); } /** Index: tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java =================================================================== --- tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java (revision 1613819) +++ tika-parsers/src/test/java/org/apache/tika/parser/mp4/MP4ParserTest.java (working copy) @@ -81,7 +81,11 @@ assertEquals("Test Genre", metadata.get(XMPDM.GENRE)); assertEquals("Test Comments", metadata.get(XMPDM.LOG_COMMENT.getName())); assertEquals("1", metadata.get(XMPDM.TRACK_NUMBER)); + assertEquals("Test Album Artist", metadata.get(XMPDM.ALBUM_ARTIST)); + assertEquals("6", metadata.get(XMPDM.DISC_NUMBER)); + assertEquals("0", metadata.get(XMPDM.COMPILATION)); + assertEquals("44100", metadata.get(XMPDM.AUDIO_SAMPLE_RATE)); //assertEquals("Stereo", metadata.get(XMPDM.AUDIO_CHANNEL_TYPE)); // TODO Extract assertEquals("M4A", metadata.get(XMPDM.AUDIO_COMPRESSOR)); Index: tika-parsers/src/test/resources/test-documents/testMP3id3v2.mp3 =================================================================== Cannot display: file marked as a binary type. svn:mime-type = application/octet-stream Index: tika-parsers/src/test/resources/test-documents/testMP3id3v24.mp3 =================================================================== Cannot display: file marked as a binary type. svn:mime-type = application/octet-stream