[FFmpeg-devel] [PATCH 01/19] avformat/webvttdec: add support for HLS WebVTT MPEG timestamp maps

[Anssi Hannula]

WebVTT subtitle files in HLS streams contain a header to synchronize the
subtitles with the MPEG TS timestamps of the HLS stream.

Add an AVOption to prefer MPEG TS style timestamps generated according
to the mapping header, if available.

Signed-off-by: Anssi Hannula <anssi.hannula at iki.fi>
---
 libavformat/webvttdec.c | 46 ++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 44 insertions(+), 2 deletions(-)

diff --git a/libavformat/webvttdec.c b/libavformat/webvttdec.c
index 0654485..1583e1b 100644
--- a/libavformat/webvttdec.c
+++ b/libavformat/webvttdec.c
@@ -35,6 +35,7 @@ typedef struct {
     const AVClass *class;
     FFDemuxSubtitlesQueue q;
     int kind;
+    int prefer_hls_mpegts_pts;
 } WebVTTContext;
 
 static int webvtt_probe(AVProbeData *p)
@@ -57,12 +58,19 @@ static int64_t read_ts(const char *s)
     return AV_NOPTS_VALUE;
 }
 
+static int64_t convert_to_hls_mpegts_ts(int64_t timestamp, int64_t offset)
+{
+     return (timestamp * 90 + offset) & ((1LL << 33) - 1);
+}
+
 static int webvtt_read_header(AVFormatContext *s)
 {
     WebVTTContext *webvtt = s->priv_data;
     AVBPrint header, cue;
     int res = 0;
     AVStream *st = avformat_new_stream(s, NULL);
+    int has_hls_timestamp_map = 0;
+    int64_t hls_ts_offset;
 
     if (!st)
         return AVERROR(ENOMEM);
@@ -90,10 +98,37 @@ static int webvtt_read_header(AVFormatContext *s)
         p = identifier = cue.str;
         pos = avio_tell(s->pb);
 
-        /* ignore header chunk */
+        /* header chunk */
         if (!strncmp(p, "\xEF\xBB\xBFWEBVTT", 9) ||
-            !strncmp(p, "WEBVTT", 6))
+            !strncmp(p, "WEBVTT", 6)) {
+
+            if (webvtt->prefer_hls_mpegts_pts) {
+                /*
+                 * WebVTT files in HLS streams contain a timestamp offset for
+                 * syncing with the main stream:
+                 *
+                 * X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:900000
+                 * (LOCAL and MPEGTS can be reversed even though HLS spec
+                 *  does not say so)
+                 */
+                char* hls_timestamp_map = strstr(p, "\nX-TIMESTAMP-MAP=");
+                if (hls_timestamp_map) {
+                    char *native_str = strstr(hls_timestamp_map, "LOCAL:");
+                    char *mpegts_str = strstr(hls_timestamp_map, "MPEGTS:");
+                    if (native_str && mpegts_str) {
+                        int64_t native_ts = read_ts(native_str + 6);
+                        int64_t mpegts_ts = strtoll(mpegts_str + 7, NULL, 10);
+
+                        if (native_ts != AV_NOPTS_VALUE) {
+                            hls_ts_offset = mpegts_ts - native_ts * 90;
+                            has_hls_timestamp_map = 1;
+                            avpriv_set_pts_info(st, 33, 1, 90000);
+                        }
+                    }
+                }
+            }
             continue;
+        }
 
         /* optional cue identifier (can be a number like in SRT or some kind of
          * chaptering id) */
@@ -124,6 +159,12 @@ static int webvtt_read_header(AVFormatContext *s)
         if ((ts_end = read_ts(p)) == AV_NOPTS_VALUE)
             break;
 
+        if (has_hls_timestamp_map) {
+            /* convert and truncate to MPEG TS timestamps */
+            ts_start = convert_to_hls_mpegts_ts(ts_start, hls_ts_offset);
+            ts_end = convert_to_hls_mpegts_ts(ts_end, hls_ts_offset);
+        }
+
         /* optional cue settings */
         p += strcspn(p, "\n\t ");
         while (*p == '\t' || *p == ' ')
@@ -199,6 +240,7 @@ static const AVOption options[] = {
         { "captions",     "WebVTT captions kind",     0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_CAPTIONS },     INT_MIN, INT_MAX, 0, "webvtt_kind" },
         { "descriptions", "WebVTT descriptions kind", 0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_DESCRIPTIONS }, INT_MIN, INT_MAX, 0, "webvtt_kind" },
         { "metadata",     "WebVTT metadata kind",     0, AV_OPT_TYPE_CONST, { .i64 = AV_DISPOSITION_METADATA },     INT_MIN, INT_MAX, 0, "webvtt_kind" },
+    { "prefer_hls_mpegts_pts", "Use WebVTT embedded HLS MPEG TS timestamps, if available.", OFFSET(prefer_hls_mpegts_pts), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, AV_OPT_FLAG_SUBTITLE_PARAM | AV_OPT_FLAG_DECODING_PARAM },
     { NULL }
 };
 
-- 
1.8.1.5