[FFmpeg-devel] [PATCH v2 7/7] avformat/audiointerleave: use a fixed frame duration for non-audio packets

[Marton Balint]

The packet durations might not be set properly which can cause the MXF muxer
to write more than one packet of a stream to an edit unit messing up the
constant byte per element index...

Also use nb_samples directly to calculate dts of audio packets because adding
packet durations might not be precise.

Signed-off-by: Marton Balint <cus at passwd.hu>
---
 libavformat/audiointerleave.c | 12 +++++++++---
 libavformat/audiointerleave.h |  3 ++-
 libavformat/gxfenc.c          |  2 +-
 libavformat/mxfenc.c          |  2 +-
 4 files changed, 13 insertions(+), 6 deletions(-)

diff --git a/libavformat/audiointerleave.c b/libavformat/audiointerleave.c
index 2e83031bd6..0643159770 100644
--- a/libavformat/audiointerleave.c
+++ b/libavformat/audiointerleave.c
@@ -40,6 +40,7 @@ void ff_audio_interleave_close(AVFormatContext *s)
 
 int ff_audio_interleave_init(AVFormatContext *s,
                              const int samples_per_frame,
+                             const int frame_duration,
                              AVRational time_base)
 {
     int i;
@@ -48,6 +49,10 @@ int ff_audio_interleave_init(AVFormatContext *s,
         av_log(s, AV_LOG_ERROR, "timebase not set for audio interleave\n");
         return AVERROR(EINVAL);
     }
+    if (!frame_duration) {
+        av_log(s, AV_LOG_ERROR, "frame_duration not set for audio interleave\n");
+        return AVERROR(EINVAL);
+    }
     for (i = 0; i < s->nb_streams; i++) {
         AVStream *st = s->streams[i];
         AudioInterleaveContext *aic = st->priv_data;
@@ -67,6 +72,8 @@ int ff_audio_interleave_init(AVFormatContext *s,
             if (!(aic->fifo = av_fifo_alloc_array(100, max_samples)))
                 return AVERROR(ENOMEM);
             aic->fifo_size = 100 * max_samples;
+        } else {
+            aic->frame_duration = frame_duration;
         }
     }
 
@@ -94,10 +101,9 @@ static int interleave_new_audio_packet(AVFormatContext *s, AVPacket *pkt,
     if (size < pkt->size)
         memset(pkt->data + size, 0, pkt->size - size);
 
-    pkt->dts = pkt->pts = aic->dts;
+    pkt->dts = pkt->pts = av_rescale_q(aic->nb_samples, st->time_base, aic->time_base);
     pkt->duration = av_rescale_q(nb_samples, st->time_base, aic->time_base);
     pkt->stream_index = stream_index;
-    aic->dts += pkt->duration;
     aic->nb_samples += nb_samples;
     aic->n++;
 
@@ -124,7 +130,7 @@ int ff_audio_rechunk_interleave(AVFormatContext *s, AVPacket *out, AVPacket *pkt
         } else {
             // rewrite pts and dts to be decoded time line position
             pkt->pts = pkt->dts = aic->dts;
-            aic->dts += pkt->duration;
+            aic->dts += aic->frame_duration;
             if ((ret = ff_interleave_add_packet(s, pkt, compare_ts)) < 0)
                 return ret;
         }
diff --git a/libavformat/audiointerleave.h b/libavformat/audiointerleave.h
index 0933310f4c..0c284dedb6 100644
--- a/libavformat/audiointerleave.h
+++ b/libavformat/audiointerleave.h
@@ -34,10 +34,11 @@ typedef struct AudioInterleaveContext {
     uint64_t dts;                 ///< current dts
     int sample_size;              ///< size of one sample all channels included
     int samples_per_frame;        ///< samples per frame if fixed, 0 otherwise
+    int frame_duration;           ///< frame duration for non-audio data
     AVRational time_base;         ///< time base of output audio packets
 } AudioInterleaveContext;
 
-int ff_audio_interleave_init(AVFormatContext *s, const int samples_per_frame, AVRational time_base);
+int ff_audio_interleave_init(AVFormatContext *s, const int samples_per_frame, const int frame_duration, AVRational time_base);
 void ff_audio_interleave_close(AVFormatContext *s);
 
 /**
diff --git a/libavformat/gxfenc.c b/libavformat/gxfenc.c
index e7536a6a7e..552cc57a3f 100644
--- a/libavformat/gxfenc.c
+++ b/libavformat/gxfenc.c
@@ -818,7 +818,7 @@ static int gxf_write_header(AVFormatContext *s)
         sc->order = s->nb_streams - st->index;
     }
 
-    if (ff_audio_interleave_init(s, GXF_samples_per_frame, (AVRational){ 1, 48000 }) < 0)
+    if (ff_audio_interleave_init(s, GXF_samples_per_frame, 2, (AVRational){ 1, 48000 }) < 0)
         return -1;
 
     if (tcr && vsc)
diff --git a/libavformat/mxfenc.c b/libavformat/mxfenc.c
index 6279ba9d6d..ac409d9ccf 100644
--- a/libavformat/mxfenc.c
+++ b/libavformat/mxfenc.c
@@ -2646,7 +2646,7 @@ static int mxf_write_header(AVFormatContext *s)
         return AVERROR(ENOMEM);
     mxf->timecode_track->index = -1;
 
-    if (ff_audio_interleave_init(s, 0, av_inv_q(mxf->tc.rate)) < 0)
+    if (ff_audio_interleave_init(s, 0, 1, av_inv_q(mxf->tc.rate)) < 0)
         return -1;
 
     return 0;
-- 
2.16.4