Handle a Theora stream with empty data packets

core/src/main/java/org/gagravarr/theora/TheoraPacketFactory.java

@@ -78,10 +78,9 @@ protected static boolean isTheoraSpecial(OggPacket packet) {
      *  instance based on the type.
      */
     public static TheoraPacket create(OggPacket packet) {
-        byte type = packet.getData()[0];
-
         // Special header types detection
         if(isTheoraSpecial(packet)) {
+            byte type = packet.getData()[0];
             switch(type) {
             case (byte)TYPE_IDENTIFICATION:
                 return new TheoraInfo(packet);

core/src/test/java/org/gagravarr/theora/TestTheoraFileRead.java

@@ -27,6 +27,7 @@
 import org.gagravarr.ogg.OggStreamIdentifier;
 import org.gagravarr.ogg.OggStreamVideoData;
 import org.gagravarr.opus.OpusAudioData;
+import org.gagravarr.skeleton.SkeletonFisbone;
 import org.gagravarr.skeleton.SkeletonStream;
 import org.gagravarr.speex.SpeexAudioData;
 import org.gagravarr.vorbis.VorbisAudioData;
@@ -52,6 +53,10 @@ private InputStream getTheoraVorbisOpusSpeexFile() throws IOException {
     }
     // TODO Finish the other test files and use them
 
+    private InputStream getTheoraEmptyPacketsFile() throws IOException {
+        return this.getClass().getResourceAsStream("/TIKA-1856-1.ogx");
+    }
+
     private TheoraFile tf;
     @Override
     protected void tearDown() throws IOException {
@@ -471,4 +476,71 @@ public void testGetAudioVisualDataBySid() throws Exception {
         assertEquals(3, speexPackets);
         assertEquals(9, vorbisPackets);
     }
+
+    /**
+     * TIKA-1856 / Issue #51
+     * If the video data packet is empty, don't break!
+     */
+    public void testEmptyPackets() throws Exception {
+        OggFile ogg = new OggFile(getTheoraEmptyPacketsFile());
+        tf = new TheoraFile(ogg);
+
+        // Check the Info
+        assertEquals("3.2.1", tf.getInfo().getVersion());
+        assertEquals(3, tf.getInfo().getMajorVersion());
+        assertEquals(2, tf.getInfo().getMinorVersion());
+        assertEquals(1, tf.getInfo().getRevisionVersion());
+
+        assertEquals(40, tf.getInfo().getFrameWidthMB());
+        assertEquals(27, tf.getInfo().getFrameHeightMB());
+        assertEquals(640, tf.getInfo().getFrameWidth());
+        assertEquals(432, tf.getInfo().getFrameHeight());
+        assertEquals(640, tf.getInfo().getPictureRegionWidth());
+        assertEquals(425, tf.getInfo().getPictureRegionHeight());
+        assertEquals(0, tf.getInfo().getPictureRegionXOffset());
+        assertEquals(5, tf.getInfo().getPictureRegionYOffset());
+
+        // Check the Comments
+        assertEquals(
+                "Xiph.Org libtheora 1.1 20090822 (Thusnelda)",
+                tf.getComments().getVendor()
+        );
+        assertEquals(2, tf.getComments().getAllComments().size());
+
+        // Check the setup
+        assertNotNull(tf.getSetup());
+
+        // Has a basic skeleton
+        assertNotNull(tf.getSkeleton());
+        assertNotNull(tf.getSkeleton().getFishead());
+        assertNotNull(tf.getSkeleton().getFisbones());
+        assertEquals(1, tf.getSkeleton().getFisbones().size());
+        SkeletonFisbone fisbone = tf.getSkeleton().getFisbones().get(0);
+        assertEquals("video/theora", fisbone.getMessageHeaders().get("Content-Type"));
+
+        // No soundtracks, video only
+        assertNotNull(tf.getSoundtracks());
+        assertNotNull(tf.getSoundtrackStreams());
+        assertEquals(0, tf.getSoundtracks().size());
+        assertEquals(0, tf.getSoundtrackStreams().size());
+
+        // Read all the video packets - shouldn't crash even on empty ones!
+        OggStreamAudioVisualData avd = null;
+        int numTheora = 0;
+        int emptyVideo = 0;
+        while ((avd = tf.getNextAudioVisualPacket()) != null) {
+            if (avd instanceof TheoraVideoData) {
+                numTheora++;
+                if (avd.getData() != null && avd.getData().length == 0) {
+                    emptyVideo++;
+                }
+            } else {
+                fail("Unexpected packet " + avd.getClass());
+            }
+        }
+
+        // Verify we found some of each
+        assertTrue("Not enough video frames, found " + numTheora, numTheora > 0);
+        assertTrue("Not enough empty video frames, found " + emptyVideo, emptyVideo > 0);
+    }
 }