Class OggParser

tika/src/main/java/org/gagravarr/tika/OggParser.java:48–155

General parser for Ogg files where we don't know what the specific kind is. We provide a detector which should help specialise the more common kinds of Ogg files, based on their streams, to their appropriate types. This just handles the rest, as best we can.

 *  appropriate types. This just handles the rest, as best we can.
 */
public class OggParser extends AbstractParser {
   private static final long serialVersionUID = -5686095376587813226L;

   protected static final MediaType OGG_GENERAL = OggDetector.OGG_GENERAL;
   protected static final MediaType OGG_AUDIO = OggDetector.OGG_AUDIO;
   protected static final MediaType OGG_VIDEO = OggDetector.OGG_VIDEO;

   protected static final MediaType KATE =
           MediaType.parse(OggStreamIdentifier.KATE.mimetype);
   
   protected static final MediaType DAALA_VIDEO =
           MediaType.parse(OggStreamIdentifier.DAALA_VIDEO.mimetype);
   protected static final MediaType DIRAC_VIDEO =
           MediaType.parse(OggStreamIdentifier.DIRAC_VIDEO.mimetype);
   protected static final MediaType OGM_VIDEO =
           MediaType.parse(OggStreamIdentifier.OGM_VIDEO.mimetype);
   protected static final MediaType OGG_UVS =
           MediaType.parse(OggStreamIdentifier.OGG_UVS.mimetype);
   protected static final MediaType OGG_YUV =
           MediaType.parse(OggStreamIdentifier.OGG_YUV.mimetype);
   protected static final MediaType OGG_RGB =
           MediaType.parse(OggStreamIdentifier.OGG_RGB.mimetype);
   protected static final MediaType OGG_PCM =
           MediaType.parse(OggStreamIdentifier.OGG_PCM.mimetype);

   private static List<MediaType> TYPES = Arrays.asList(new MediaType[] {
         // General ones, which didn't fit a specific pattern (eg 2*Opus+2*Vorbis,
         //  Unknown audio codec, unknown video codec etc). We can't do much
         //  for these
         OGG_GENERAL, OGG_AUDIO, OGG_VIDEO,
         // Ones we lack a proper parser for, but may be able to support
         //  better in the future with a dedicated parser
         DAALA_VIDEO, DIRAC_VIDEO, OGM_VIDEO,
         OGG_UVS, OGG_YUV,  OGG_RGB, OGG_PCM,
         KATE
   });

   public Set<MediaType> getSupportedTypes(ParseContext context) {
      return new HashSet<MediaType>(TYPES);
   }

   public void parse(
         InputStream stream, ContentHandler handler,
         Metadata metadata, ParseContext context)
         throws IOException, TikaException, SAXException {
      // Process the file straight through once
      OggFile ogg = new OggFile(stream);
      
      // To track the streams we find
      Map<OggStreamType, Integer> streams = 
              new HashMap<OggStreamType, Integer>();
      Map<OggStreamType.Kind, Integer> streamKinds =
              new HashMap<OggStreamType.Kind, Integer>();
      List<Integer> sids = new ArrayList<Integer>();
      int totalStreams = 0;
      
      // Start
      XHTMLContentHandler xhtml = new XHTMLContentHandler(handler, metadata);