mov: Add option to keep exact packet sequence after seeking

The current behavior may produce a different sequence of packets
after seeking, compared to demuxing linearly from the beginning.
This is because the MOV demuxer seeks in each stream individually,
based on timestamp, which may set each stream at a slightly different
position than if the file would have been read sequentially.

This makes implementing certain operations, such as segmenting,
quite hard, and slower than need be.

Therefore, add an option which retains the same packet sequence
after seeking, as when a file is demuxed linearly.

1 files changed, 33 insertions, 9 deletions

diff --git a/libavformat/mov.c b/libavformat/mov.c
index 45c843f53e..76ce9ee217 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -3468,8 +3468,8 @@ static int mov_seek_stream(AVFormatContext *s, AVStream *st, int64_t timestamp,
 
 static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_time, int flags)
 {
+    MOVContext *mc = s->priv_data;
     AVStream *st;
-    int64_t seek_timestamp, timestamp;
     int sample;
     int i;
 
@@ -3483,16 +3483,36 @@ static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_ti
     if (sample < 0)
         return sample;
 
-    /* adjust seek timestamp to found sample timestamp */
-    seek_timestamp = st->index_entries[sample].timestamp;
+    if (mc->seek_individually) {
+        /* adjust seek timestamp to found sample timestamp */
+        int64_t seek_timestamp = st->index_entries[sample].timestamp;
 
-    for (i = 0; i < s->nb_streams; i++) {
-        st = s->streams[i];
-        if (stream_index == i)
-            continue;
+        for (i = 0; i < s->nb_streams; i++) {
+            int64_t timestamp;
+            st = s->streams[i];
+            if (stream_index == i)
+                continue;
 
-        timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
-        mov_seek_stream(s, st, timestamp, flags);
+            timestamp = av_rescale_q(seek_timestamp, s->streams[stream_index]->time_base, st->time_base);
+            mov_seek_stream(s, st, timestamp, flags);
+        }
+    } else {
+        for (i = 0; i < s->nb_streams; i++) {
+            MOVStreamContext *sc;
+            st = s->streams[i];
+            sc = st->priv_data;
+            sc->current_sample = 0;
+        }
+        while (1) {
+            MOVStreamContext *sc;
+            AVIndexEntry *entry = mov_find_next_sample(s, &st);
+            if (!entry)
+                return AVERROR_INVALIDDATA;
+            sc = st->priv_data;
+            if (sc->ffindex == stream_index && sc->current_sample == sample)
+                break;
+            sc->current_sample++;
+        }
     }
     return 0;
 }
@@ -3500,6 +3520,10 @@ static int mov_read_seek(AVFormatContext *s, int stream_index, int64_t sample_ti
 #define OFFSET(x) offsetof(MOVContext, x)
 #define FLAGS AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
 static const AVOption mov_options[] = {
+    {"seek_streams_individually",
+        "Seek each stream individually to the to the closest point",
+        OFFSET(seek_individually), AV_OPT_TYPE_INT, { .i64 = 1 },
+        0, 1, FLAGS},
     { "export_all", "Export unrecognized metadata entries", OFFSET(export_all),
         AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 1, .flags = FLAGS },
     { "export_xmp", "Export full XMP metadata", OFFSET(export_xmp),