Patch force dts monotonicity branch (#6)

* patch for force_dts_monotonicity option to enforce continuity timestamps * update travis
loomhq · Feb 7, 2022 · 95931bd · 95931bd
1 parent a59f59f
commit 95931bd
Show file tree

Hide file tree

Showing 4 changed files with 66 additions and 6 deletions.
diff --git a/.travis.yml b/.travis.yml
@@ -2,7 +2,6 @@ language: c
 sudo: false
 os:
   - linux
-  - osx
 addons:
   apt:
     packages:
@@ -12,6 +11,9 @@ compiler:
   - clang
   - gcc
 matrix:
+    include:
+        - os: osx
+          osx_image: xcode13.2
     exclude:
         - os: osx
           compiler: gcc

diff --git a/fftools/ffmpeg.c b/fftools/ffmpeg.c
@@ -4398,10 +4398,14 @@ static int process_input(int file_index)
         pkt.dts *= ist->ts_scale;
 
     pkt_dts = av_rescale_q_rnd(pkt.dts, ist->st->time_base, AV_TIME_BASE_Q, AV_ROUND_NEAR_INF|AV_ROUND_PASS_MINMAX);
-    if ((ist->dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO ||
-         ist->dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) &&
-        pkt_dts != AV_NOPTS_VALUE && ist->next_dts == AV_NOPTS_VALUE && !copy_ts
-        && (is->iformat->flags & AVFMT_TS_DISCONT) && ifile->last_ts != AV_NOPTS_VALUE) {
+    if ((ist->dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO || ist->dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) &&
+        pkt_dts != AV_NOPTS_VALUE &&
+        ist->next_dts == AV_NOPTS_VALUE &&
+        !copy_ts &&
+        (is->iformat->flags & AVFMT_TS_DISCONT) &&
+        ifile->last_ts != AV_NOPTS_VALUE &&
+        !force_dts_monotonicity
+    ) {
         int64_t delta   = pkt_dts - ifile->last_ts;
         if (delta < -1LL*dts_delta_threshold*AV_TIME_BASE ||
             delta >  1LL*dts_delta_threshold*AV_TIME_BASE){
@@ -4439,7 +4443,9 @@ static int process_input(int file_index)
     if ((ist->dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO ||
          ist->dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO) &&
          pkt_dts != AV_NOPTS_VALUE && ist->next_dts != AV_NOPTS_VALUE &&
-        !disable_discontinuity_correction) {
+         !disable_discontinuity_correction &&
+         !force_dts_monotonicity
+        ) {
         int64_t delta   = pkt_dts - ist->next_dts;
         if (is->iformat->flags & AVFMT_TS_DISCONT) {
             if (delta < -1LL*dts_delta_threshold*AV_TIME_BASE ||
@@ -4477,6 +4483,43 @@ static int process_input(int file_index)
     if (pkt.dts != AV_NOPTS_VALUE)
         ifile->last_ts = av_rescale_q(pkt.dts, ist->st->time_base, AV_TIME_BASE_Q);
 
+    if (force_dts_monotonicity &&
+        (pkt.pts != AV_NOPTS_VALUE || pkt.dts != AV_NOPTS_VALUE) &&
+        (ist->dec_ctx->codec_type == AVMEDIA_TYPE_VIDEO || ist->dec_ctx->codec_type == AVMEDIA_TYPE_AUDIO)
+    ) {
+        int64_t ff_pts_error = 0;
+        int64_t ff_dts_error = 0;
+        int64_t ff_dts_threshold = av_rescale_q(dts_monotonicity_threshold, AV_TIME_BASE_Q, ist->st->time_base);
+
+        // adjust the incoming packet by the accumulated monotonicity error
+        if (pkt.pts != AV_NOPTS_VALUE) {
+            pkt.pts += ifile->ff_timestamp_monotonicity_offset;
+            if (ist->next_pts != AV_NOPTS_VALUE) {
+                ff_pts_error = av_rescale_q(ist->next_pts, AV_TIME_BASE_Q, ist->st->time_base) - pkt.pts;
+            }
+        }
+        if (pkt.dts != AV_NOPTS_VALUE) {
+            pkt.dts += ifile->ff_timestamp_monotonicity_offset;
+            if (ist->next_dts != AV_NOPTS_VALUE) {
+                ff_dts_error = av_rescale_q(ist->next_dts, AV_TIME_BASE_Q, ist->st->time_base) - pkt.dts;
+            }
+        }
+
+        if(ff_dts_error > 0 || ff_dts_error < (-ff_dts_threshold) || ff_pts_error < (-ff_dts_threshold)) {
+            if(pkt.dts == AV_NOPTS_VALUE /*  || ist->next_dts != AV_NOPTS_VALUE */ ) {
+                pkt.pts += ff_pts_error;
+                ifile->ff_timestamp_monotonicity_offset += ff_pts_error;
+                av_log(is, AV_LOG_INFO, "Incoming PTS error %"PRId64", offsetting subsequent timestamps by %"PRId64" to correct\n", ff_pts_error, ifile->ff_timestamp_monotonicity_offset);
+            }
+            else {
+                pkt.pts += ff_dts_error;
+                pkt.dts += ff_dts_error;
+                ifile->ff_timestamp_monotonicity_offset += ff_dts_error;
+                av_log(is, AV_LOG_INFO, "Incoming DTS error %"PRId64", offsetting subsequent timestamps by %"PRId64" to correct\n", ff_dts_error, ifile->ff_timestamp_monotonicity_offset);
+            }
+        }
+    }
+
     if (debug_ts) {
         av_log(NULL, AV_LOG_INFO, "demuxer+ffmpeg -> ist_index:%d type:%s pkt_pts:%s pkt_pts_time:%s pkt_dts:%s pkt_dts_time:%s off:%s off_time:%s\n",
                ifile->ist_index + pkt.stream_index, av_get_media_type_string(ist->dec_ctx->codec_type),

diff --git a/fftools/ffmpeg.h b/fftools/ffmpeg.h
@@ -419,6 +419,10 @@ typedef struct InputFile {
     int joined;                 /* the thread has been joined */
     int thread_queue_size;      /* maximum number of queued packets */
 #endif
+
+    // A value added to inbound timestamps to prevent them from going "backward" in cases such as HLS discontinuities
+    int64_t ff_timestamp_monotonicity_offset;
+
 } InputFile;
 
 enum forced_keyframes_const {
@@ -584,6 +588,9 @@ extern float audio_drift_threshold;
 extern float dts_delta_threshold;
 extern float dts_error_threshold;
 
+extern int dts_monotonicity_threshold;
+extern int force_dts_monotonicity;
+
 extern int audio_volume;
 extern int audio_sync_method;
 extern int video_sync_method;

diff --git a/fftools/ffmpeg_opt.c b/fftools/ffmpeg_opt.c
@@ -148,6 +148,9 @@ float audio_drift_threshold = 0.1;
 float dts_delta_threshold   = 10;
 float dts_error_threshold   = 3600*30;
 
+int dts_monotonicity_threshold = AV_TIME_BASE;
+int force_dts_monotonicity = 0;
+
 int audio_volume      = 256;
 int audio_sync_method = 0;
 int video_sync_method = VSYNC_AUTO;
@@ -1262,6 +1265,7 @@ static int open_input_file(OptionsContext *o, const char *filename)
     f->recording_time = o->recording_time;
     f->input_ts_offset = o->input_ts_offset;
     f->ts_offset  = o->input_ts_offset - (copy_ts ? (start_at_zero && ic->start_time != AV_NOPTS_VALUE ? ic->start_time : 0) : timestamp);
+    f->ff_timestamp_monotonicity_offset = 0;
     f->nb_streams = ic->nb_streams;
     f->rate_emu   = o->rate_emu;
     f->accurate_seek = o->accurate_seek;
@@ -3499,6 +3503,10 @@ const OptionDef options[] = {
     { "apad",           OPT_STRING | HAS_ARG | OPT_SPEC |
                         OPT_OUTPUT,                                  { .off = OFFSET(apad) },
         "audio pad", "" },
+    { "force_dts_monotonicity", OPT_BOOL | OPT_EXPERT,               { &force_dts_monotonicity },
+        "correct dts monotonicity errors" },
+    { "dts_monotonicity_threshold",    HAS_ARG | OPT_INT | OPT_EXPERT,{ &dts_monotonicity_threshold },
+        "dts correction threshold for forward jumps", "microseconds" },
     { "dts_delta_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT,       { &dts_delta_threshold },
         "timestamp discontinuity delta threshold", "threshold" },
     { "dts_error_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT,       { &dts_error_threshold },