movdec: Interpret the ismv specific sdtp atom

diff mbox

diff --git a/libavformat/isom.h b/libavformat/isom.h
index e74a552..4d4e307 100644
--- a/libavformat/isom.h
+++ b/libavformat/isom.h
@@ -77,6 +77,7 @@  typedef struct {
     unsigned duration;
     unsigned size;
     unsigned flags;
+    unsigned trun_entries;
 } MOVFragment;
 
 typedef struct {
@@ -138,6 +139,7 @@  typedef struct MOVContext {
     DVDemuxContext *dv_demux;
     AVFormatContext *dv_fctx;
     int isom;             ///< 1 if file is ISO Media (mp4/3gp)
+    int ismv;             ///< 1 if file is ISMV/ISMA
     MOVFragment fragment; ///< current fragment in moof atom
     MOVTrackExt *trex_data;
     unsigned trex_count;
diff --git a/libavformat/mov.c b/libavformat/mov.c
index 690cd1e..9c0a143 100644
--- a/libavformat/mov.c
+++ b/libavformat/mov.c
@@ -664,6 +664,8 @@  static int mov_read_ftyp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
     avio_read(pb, type, 4);
     if (strcmp(type, "qt  "))
         c->isom = 1;
+    if (!strcmp(type, "isml"))
+        c->ismv = 1;
     av_log(c->fc, AV_LOG_DEBUG, "ISO: File Type Major Brand: %.4s\n",(char *)&type);
     av_dict_set(&c->fc->metadata, "major_brand", type, 0);
     minor_ver = avio_rb32(pb); /* minor version */
@@ -2292,10 +2294,54 @@  static int mov_read_trun(MOVContext *c, AVIOContext *pb, MOVAtom atom)
         sc->data_size += sample_size;
     }
     frag->moof_offset = offset;
+    frag->trun_entries = entries;
     st->duration = sc->track_end = dts + sc->time_offset;
     return 0;
 }
 
+static int mov_read_sdtp(MOVContext *c, AVIOContext *pb, MOVAtom atom)
+{
+    MOVFragment *frag = &c->fragment;
+    AVStream *st = NULL;
+    int i;
+
+    /* Interpret the ismv specific sdtp atom. After the version+flags,
+     * there's one byte per sample in the trun. The bytes seem to be 0x28
+     * for audio streams, 0x24 for video keyframes and 0x14 for video
+     * nonkeyframes.
+     * The sample flags seem to have a different meaning in these files, too,
+     * compared to the iso media spec. For ismv, the flags normally are
+     * 0x8002 for audio samples, 0x4002 for video keyframes and 0x4001 for
+     * nonkeyframes. However, some files don't have sample flags at all,
+     * others don't indicate any difference between keyframes and
+     * nonkeyframes in the flags. By parsing the sdtp, we get keyframes
+     * marked properly. */
+    if (!c->ismv)
+        return 0;
+
+    for (i = 0; i < c->fc->nb_streams && !st; i++)
+        if (c->fc->streams[i]->id == frag->track_id)
+            st = c->fc->streams[i];
+    if (!st) {
+        av_log(c->fc, AV_LOG_ERROR,
+               "could not find corresponding track id %d\n", frag->track_id);
+        return AVERROR_INVALIDDATA;
+    }
+    avio_r8(pb); /* version */
+    avio_rb24(pb); /* flags */
+
+    if (frag->trun_entries == atom.size - 4 &&
+        st->nb_index_entries >= frag->trun_entries) {
+        for (i = 0; i < frag->trun_entries; i++) {
+            int entry = st->nb_index_entries - frag->trun_entries + i;
+            AVIndexEntry *ie = &st->index_entries[entry];
+            if (avio_r8(pb) & 0x20)
+                ie->flags |= AVINDEX_KEYFRAME;
+        }
+    }
+    return 0;
+}
+
 /* this atom should be null (from specs), but some buggy files put the 'moov' atom inside it... */
 /* like the files created with Adobe Premiere 5.0, for samples see */
 /* http://graphics.tudelft.nl/~wouter/publications/soundtests/ */
@@ -2455,6 +2501,7 @@  static const MOVParseTableEntry mov_default_parse_table[] = {
 { MKTAG('c','h','a','p'), mov_read_chap },
 { MKTAG('t','r','e','x'), mov_read_trex },
 { MKTAG('t','r','u','n'), mov_read_trun },
+{ MKTAG('s','d','t','p'), mov_read_sdtp },
 { MKTAG('u','d','t','a'), mov_read_default },
 { MKTAG('w','a','v','e'), mov_read_wave },
 { MKTAG('e','s','d','s'), mov_read_esds },