Use deinterleavers for demangling audio packets in RealMedia.

Unlike other containers RealMedia stores its audio packets in scrambled form,
with interleaver ID preceeding audio codec ID. Currently deinterleaving
decision is tied to the codec while it's possible to have non-default
deinterleaver with audio codec (like Int0 deinterleaver instead of specific
one for Sipro).

Signed-off-by: Anton Khirnov <anton@khirnov.net>

1 files changed, 34 insertions, 12 deletions

diff --git a/libavformat/rmdec.c b/libavformat/rmdec.c
index 7cf5720fe0..02ff7e93f0 100644
--- a/libavformat/rmdec.c
+++ b/libavformat/rmdec.c
@@ -26,6 +26,13 @@
 #include "riff.h"
 #include "rm.h"
 
+#define DEINT_ID_GENR MKTAG('g', 'e', 'n', 'r') ///< interleaving for Cooker/Atrac
+#define DEINT_ID_INT0 MKTAG('I', 'n', 't', '0') ///< no interleaving needed
+#define DEINT_ID_INT4 MKTAG('I', 'n', 't', '4') ///< interleaving for 28.8
+#define DEINT_ID_SIPR MKTAG('s', 'i', 'p', 'r') ///< interleaving for Sipro
+#define DEINT_ID_VBRF MKTAG('v', 'b', 'r', 'f') ///< VBR case for AAC
+#define DEINT_ID_VBRS MKTAG('v', 'b', 'r', 's') ///< VBR case for AAC
+
 struct RMStream {
     AVPacket pkt;      ///< place to store merged video frame / reordered audio data
     int videobufsize;  ///< current assembled frame size
@@ -39,6 +46,7 @@ struct RMStream {
     int sub_packet_size, sub_packet_h, coded_framesize; ///< Descrambling parameters from container
     int audio_framesize; /// Audio frame size from container
     int sub_packet_lengths[16]; /// Length of each subpacket
+    int32_t deint_id;  ///< deinterleaver used in audio stream
 };
 
 typedef struct {
@@ -147,6 +155,7 @@ static int rm_read_audio_stream_info(AVFormatContext *s, AVIOContext *pb,
         st->codec->channels = 1;
         st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
         st->codec->codec_id = CODEC_ID_RA_144;
+        ast->deint_id = DEINT_ID_INT0;
     } else {
         int flavor, sub_packet_h, coded_framesize, sub_packet_size;
         int codecdata_length;
@@ -172,17 +181,31 @@ static int rm_read_audio_stream_info(AVFormatContext *s, AVIOContext *pb,
         avio_rb32(pb);
         st->codec->channels = avio_rb16(pb);
         if (version == 5) {
-            avio_rb32(pb);
+            ast->deint_id = avio_rl32(pb);
             avio_read(pb, buf, 4);
             buf[4] = 0;
         } else {
             get_str8(pb, buf, sizeof(buf)); /* desc */
+            ast->deint_id = AV_RL32(buf);
             get_str8(pb, buf, sizeof(buf)); /* desc */
         }
         st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
         st->codec->codec_tag  = AV_RL32(buf);
         st->codec->codec_id   = ff_codec_get_id(ff_rm_codec_tags,
                                                 st->codec->codec_tag);
+
+        switch (ast->deint_id) {
+        case DEINT_ID_GENR:
+        case DEINT_ID_INT0:
+        case DEINT_ID_INT4:
+        case DEINT_ID_SIPR:
+        case DEINT_ID_VBRS:
+        case DEINT_ID_VBRF:
+            break;
+        default:
+            av_log(NULL,0,"Unknown interleaver %X\n", ast->deint_id);
+            return AVERROR_INVALIDDATA;
+        }
         switch (st->codec->codec_id) {
         case CODEC_ID_AC3:
             st->need_parsing = AVSTREAM_PARSE_FULL;
@@ -704,10 +727,9 @@ ff_rm_parse_packet (AVFormatContext *s, AVIOContext *pb,
         if(rm_assemble_video_frame(s, pb, rm, ast, pkt, len, seq, &timestamp))
             return -1; //got partial frame
     } else if (st->codec->codec_type == AVMEDIA_TYPE_AUDIO) {
-        if ((st->codec->codec_id == CODEC_ID_RA_288) ||
-            (st->codec->codec_id == CODEC_ID_COOK) ||
-            (st->codec->codec_id == CODEC_ID_ATRAC3) ||
-            (st->codec->codec_id == CODEC_ID_SIPR)) {
+        if ((ast->deint_id == DEINT_ID_GENR) ||
+            (ast->deint_id == DEINT_ID_INT4) ||
+            (ast->deint_id == DEINT_ID_SIPR)) {
             int x;
             int sps = ast->sub_packet_size;
             int cfs = ast->coded_framesize;
@@ -720,30 +742,30 @@ ff_rm_parse_packet (AVFormatContext *s, AVIOContext *pb,
             if (!y)
                 ast->audiotimestamp = timestamp;
 
-            switch(st->codec->codec_id) {
-                case CODEC_ID_RA_288:
+            switch (ast->deint_id) {
+                case DEINT_ID_INT4:
                     for (x = 0; x < h/2; x++)
                         avio_read(pb, ast->pkt.data+x*2*w+y*cfs, cfs);
                     break;
-                case CODEC_ID_ATRAC3:
-                case CODEC_ID_COOK:
+                case DEINT_ID_GENR:
                     for (x = 0; x < w/sps; x++)
                         avio_read(pb, ast->pkt.data+sps*(h*x+((h+1)/2)*(y&1)+(y>>1)), sps);
                     break;
-                case CODEC_ID_SIPR:
+                case DEINT_ID_SIPR:
                     avio_read(pb, ast->pkt.data + y * w, w);
                     break;
             }
 
             if (++(ast->sub_packet_cnt) < h)
                 return -1;
-            if (st->codec->codec_id == CODEC_ID_SIPR)
+            if (ast->deint_id == DEINT_ID_SIPR)
                 ff_rm_reorder_sipr_data(ast->pkt.data, h, w);
 
              ast->sub_packet_cnt = 0;
              rm->audio_stream_num = st->index;
              rm->audio_pkt_cnt = h * w / st->codec->block_align;
-        } else if (st->codec->codec_id == CODEC_ID_AAC) {
+        } else if ((ast->deint_id == DEINT_ID_VBRF) ||
+                   (ast->deint_id == DEINT_ID_VBRS)) {
             int x;
             rm->audio_stream_num = st->index;
             ast->sub_packet_cnt = (avio_rb16(pb) & 0xf0) >> 4;