oggdec: output correct timestamps for Vorbis

Takes encoder delay into account by comparing first the coded page
duration with the calculated page duration. Handles last packet duration
if needed, also by comparing coded duration with calculated duration.
Also does better handling of timestamp generation for packets in the
first page for streamed ogg files where the start time is not
necessarily zero.

2 files changed, 82 insertions, 3 deletions

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 0c459f690f..78d9d5848c 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -569,7 +569,7 @@ OBJS-$(CONFIG_MPEGTS_MUXER)            += mpegvideo.o mpeg4audio.o
 OBJS-$(CONFIG_MPEGTS_DEMUXER)          += mpeg4audio.o mpegaudiodata.o
 OBJS-$(CONFIG_NUT_MUXER)               += mpegaudiodata.o
 OBJS-$(CONFIG_OGG_DEMUXER)             += flacdec.o flacdata.o flac.o \
-                                          dirac.o mpeg12data.o
+                                          dirac.o mpeg12data.o vorbis_parser.o
 OBJS-$(CONFIG_OGG_MUXER)               += xiph.o flacdec.o flacdata.o flac.o
 OBJS-$(CONFIG_RTP_MUXER)               += mpeg4audio.o mpegvideo.o xiph.o
 OBJS-$(CONFIG_SPDIF_DEMUXER)           += aacadtsdec.o mpeg4audio.o
diff --git a/libavformat/oggparsevorbis.c b/libavformat/oggparsevorbis.c
index b9d9f575e0..f72fd26c4c 100644
--- a/libavformat/oggparsevorbis.c
+++ b/libavformat/oggparsevorbis.c
@@ -28,6 +28,7 @@
 #include "libavutil/dict.h"
 #include "libavcodec/get_bits.h"
 #include "libavcodec/bytestream.h"
+#include "libavcodec/vorbis_parser.h"
 #include "avformat.h"
 #include "internal.h"
 #include "oggdec.h"
@@ -162,6 +163,9 @@ ff_vorbis_comment(AVFormatContext * as, AVDictionary **m, const uint8_t *buf, in
 struct oggvorbis_private {
     unsigned int len[3];
     unsigned char *packet[3];
+    VorbisParseContext vp;
+    int64_t final_pts;
+    int final_duration;
 };
 
 
@@ -251,7 +255,6 @@ vorbis_header (AVFormatContext * s, int idx)
 
         st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
         st->codec->codec_id = CODEC_ID_VORBIS;
-        st->need_parsing = AVSTREAM_PARSE_HEADERS;
 
         if (srate > 0) {
             st->codec->sample_rate = srate;
@@ -269,15 +272,91 @@ vorbis_header (AVFormatContext * s, int idx)
             }
         }
     } else {
+        int ret;
         st->codec->extradata_size =
             fixup_vorbis_headers(s, priv, &st->codec->extradata);
+        if ((ret = avpriv_vorbis_parse_extradata(st->codec, &priv->vp))) {
+            av_freep(&st->codec->extradata);
+            st->codec->extradata_size = 0;
+            return ret;
+        }
     }
 
     return 1;
 }
 
+static int vorbis_packet(AVFormatContext *s, int idx)
+{
+    struct ogg *ogg = s->priv_data;
+    struct ogg_stream *os = ogg->streams + idx;
+    struct oggvorbis_private *priv = os->private;
+    int duration;
+
+    /* first packet handling
+       here we parse the duration of each packet in the first page and compare
+       the total duration to the page granule to find the encoder delay and
+       set the first timestamp */
+    if (!os->lastpts) {
+        int seg;
+        uint8_t *last_pkt = os->buf + os->pstart;
+        uint8_t *next_pkt = last_pkt;
+        int first_duration = 0;
+
+        avpriv_vorbis_parse_reset(&priv->vp);
+        duration = 0;
+        for (seg = 0; seg < os->nsegs; seg++) {
+            if (os->segments[seg] < 255) {
+                int d = avpriv_vorbis_parse_frame(&priv->vp, last_pkt, 1);
+                if (d < 0) {
+                    duration = os->granule;
+                    break;
+                }
+                if (!duration)
+                    first_duration = d;
+                duration += d;
+                last_pkt = next_pkt + os->segments[seg];
+            }
+            next_pkt += os->segments[seg];
+        }
+        os->lastpts = os->lastdts   = os->granule - duration;
+        s->streams[idx]->start_time = os->lastpts + first_duration;
+        if (s->streams[idx]->duration)
+            s->streams[idx]->duration -= s->streams[idx]->start_time;
+        s->streams[idx]->cur_dts    = AV_NOPTS_VALUE;
+        priv->final_pts             = AV_NOPTS_VALUE;
+        avpriv_vorbis_parse_reset(&priv->vp);
+    }
+
+    /* parse packet duration */
+    if (os->psize > 0) {
+        duration = avpriv_vorbis_parse_frame(&priv->vp, os->buf + os->pstart, 1);
+        if (duration <= 0) {
+            os->pflags |= AV_PKT_FLAG_CORRUPT;
+            return 0;
+        }
+        os->pduration = duration;
+    }
+
+    /* final packet handling
+       here we save the pts of the first packet in the final page, sum up all
+       packet durations in the final page except for the last one, and compare
+       to the page granule to find the duration of the final packet */
+    if (os->flags & OGG_FLAG_EOS) {
+        if (os->lastpts != AV_NOPTS_VALUE) {
+            priv->final_pts = os->lastpts;
+            priv->final_duration = 0;
+        }
+        if (os->segp == os->nsegs)
+            os->pduration = os->granule - priv->final_pts - priv->final_duration;
+        priv->final_duration += os->pduration;
+    }
+
+    return 0;
+}
+
 const struct ogg_codec ff_vorbis_codec = {
     .magic = "\001vorbis",
     .magicsize = 7,
-    .header = vorbis_header
+    .header = vorbis_header,
+    .packet = vorbis_packet,
 };