vd_lavc: feed A53_CC side data packets into the demuxer for eia_608 decoding

demux/demux.c

@@ -160,6 +160,10 @@ struct demux_stream {
     int64_t last_pos;
     struct demux_packet *head;
     struct demux_packet *tail;
+
+    // for closed captions (demuxer_feed_caption)
+    struct sh_stream *cc;
+
 };
 
 // Return "a", or if that is NOPTS, return "def".
@@ -361,6 +365,27 @@ const char *stream_type_name(enum stream_type type)
     }
 }
 
+void demuxer_feed_caption(struct sh_stream *stream, demux_packet_t *dp)
+{
+    struct demuxer *demuxer = stream->ds->in->d_thread;
+    struct sh_stream *sh = stream->ds->cc;
+
+    if (!sh) {
+        sh = demux_alloc_sh_stream(STREAM_SUB);
+        if (!sh)
+            return;
+        sh->codec->codec = "eia_608";
+        stream->ds->cc = sh;
+        demux_add_sh_stream(demuxer, sh);
+    }
+
+    if (demux_stream_is_selected(sh)) {
+        dp->pts = MP_ADD_PTS(dp->pts, -stream->ds->in->ts_offset);
+        dp->dts = MP_ADD_PTS(dp->dts, -stream->ds->in->ts_offset);
+        demux_add_packet(sh, dp);
+    }
+}
+
 void demux_add_packet(struct sh_stream *stream, demux_packet_t *dp)
 {
     struct demux_stream *ds = stream ? stream->ds : NULL;

demux/demux.h

@@ -237,6 +237,7 @@ void free_demuxer(struct demuxer *demuxer);
 void free_demuxer_and_stream(struct demuxer *demuxer);
 
 void demux_add_packet(struct sh_stream *stream, demux_packet_t *dp);
+void demuxer_feed_caption(struct sh_stream *stream, demux_packet_t *dp);
 
 struct demux_packet *demux_read_packet(struct sh_stream *sh);
 int demux_read_packet_async(struct sh_stream *sh, struct demux_packet **out_pkt);

sub/sd_ass.c

@@ -51,6 +51,7 @@ struct sd_ass_priv {
     double sub_speed, video_fps, frame_fps;
     int64_t *seen_packets;
     int num_seen_packets;
+    bool duration_unknown;
 };
 
 static void mangle_colors(struct sd *sd, struct sub_bitmaps *parts);
@@ -176,6 +177,9 @@ static int init(struct sd *sd)
             return -1;
         extradata = lavc_conv_get_extradata(ctx->converter);
         extradata_size = extradata ? strlen(extradata) : 0;
+
+        if (strcmp(sd->codec->codec, "eia_608") == 0)
+            ctx->duration_unknown = 1;
     }
 
     ctx->ass_library = mp_ass_init(sd->global, sd->log);
@@ -239,11 +243,18 @@ static void decode(struct sd *sd, struct demux_packet *packet)
     struct sd_ass_priv *ctx = sd->priv;
     ASS_Track *track = ctx->ass_track;
     if (ctx->converter) {
-        if (!sd->opts->sub_clear_on_seek && check_packet_seen(sd, packet->pos))
+        if (!sd->opts->sub_clear_on_seek && packet->pos >= 0 &&
+            check_packet_seen(sd, packet->pos))
             return;
         char **r = lavc_conv_decode(ctx->converter, packet);
         for (int n = 0; r && r[n]; n++)
             ass_process_data(track, r[n], strlen(r[n]));
+        if (ctx->duration_unknown) {
+            for (int n = 0; n < track->n_events - 1; n++) {
+                track->events[n].Duration = track->events[n + 1].Start -
+                                            track->events[n].Start;
+            }
+        }
     } else {
         // Note that for this packet format, libass has an internal mechanism
         // for discarding duplicate (already seen) packets.
@@ -426,9 +437,12 @@ static void get_bitmaps(struct sd *sd, struct mp_osd_res dim, double pts,
     } else {
         ass_set_storage_size(renderer, 0, 0);
     }
+    long long ts = find_timestamp(sd, pts);
+    if (ctx->duration_unknown && pts != MP_NOPTS_VALUE) {
+        mp_ass_flush_old_events(track, ts);
+    }
     if (no_ass)
         fill_plaintext(sd, pts);
-    long long ts = find_timestamp(sd, pts);
     mp_ass_render_frame(renderer, track, ts, &ctx->parts, res);
     talloc_steal(ctx, ctx->parts);
 

video/decode/vd_lavc.c

@@ -41,6 +41,7 @@
 #include "video/img_format.h"
 #include "video/filter/vf.h"
 #include "video/decode/dec_video.h"
+#include "demux/demux.h"
 #include "demux/stheader.h"
 #include "demux/packet.h"
 #include "video/csputils.h"
@@ -680,6 +681,16 @@ static void decode(struct dec_video *vd, struct demux_packet *packet,
     vd->codec_pts = mp_pts_from_av(ctx->pic->pkt_pts, NULL);
     vd->codec_dts = mp_pts_from_av(ctx->pic->pkt_dts, NULL);
 
+    AVFrameSideData *sd = NULL;
+    sd = av_frame_get_side_data(ctx->pic, AV_FRAME_DATA_A53_CC);
+    if (sd) {
+        struct demux_packet *cc = new_demux_packet_from(sd->data, sd->size);
+        cc->pts = vd->codec_pts;
+        cc->dts = vd->codec_dts;
+        cc->pos = -1;
+        demuxer_feed_caption(vd->header, cc);
+    }
+
     struct mp_image *mpi = mp_image_from_av_frame(ctx->pic);
     av_frame_unref(ctx->pic);
     if (!mpi)