avformat/asfenc: write group_mutual_exclusion_objects for audio on multiple languages

Improves streaming compatibility with Windows Media Services. Also tested for
compatilbility in Windows Media Player, Windows Media ASF Viewer and VLC.

This version of the patch only writes exclusion among audio streams, therefore
choosing a subtitle language should be possible independently of audio language.

Signed-off-by: Marton Balint <cus@passwd.hu>

libavformat/asf.c

@@ -147,6 +147,14 @@ const ff_asf_guid ff_asf_extended_stream_properties_object = {
     0xcb, 0xa5, 0xe6, 0x14, 0x72, 0xc6, 0x32, 0x43, 0x83, 0x99, 0xa9, 0x69, 0x52, 0x06, 0x5b, 0x5a
 };
 
+const ff_asf_guid ff_asf_group_mutual_exclusion_object = {
+    0x40, 0x5a, 0x46, 0xd1, 0x79, 0x5a, 0x38, 0x43, 0xb7, 0x1b, 0xe3, 0x6b, 0x8f, 0xd6, 0xc2, 0x49
+};
+
+const ff_asf_guid ff_asf_mutex_language = {
+    0x00, 0x2a, 0xe2, 0xd6, 0xda, 0x35, 0xd1, 0x11, 0x90, 0x34, 0x00, 0xa0, 0xc9, 0x03, 0x49, 0xbe
+};
+
 /* List of official tags at http://msdn.microsoft.com/en-us/library/dd743066(VS.85).aspx */
 const AVMetadataConv ff_asf_metadata_conv[] = {
     { "WM/AlbumArtist",          "album_artist"     },

libavformat/asf.h

@@ -100,6 +100,8 @@ extern const ff_asf_guid ff_asf_content_encryption;
 extern const ff_asf_guid ff_asf_ext_content_encryption;
 extern const ff_asf_guid ff_asf_digital_signature;
 extern const ff_asf_guid ff_asf_extended_stream_properties_object;
+extern const ff_asf_guid ff_asf_group_mutual_exclusion_object;
+extern const ff_asf_guid ff_asf_mutex_language;
 
 extern const AVMetadataConv ff_asf_metadata_conv[];
 

libavformat/asfenc.c

@@ -396,6 +396,7 @@ static int asf_write_header1(AVFormatContext *s, int64_t file_size,
     int64_t header_offset, cur_pos, hpos;
     int bit_rate;
     int64_t duration;
+    int audio_language_counts[128] = { 0 };
 
     ff_metadata_conv(&s->metadata, ff_asf_metadata_conv, NULL);
 
@@ -444,6 +445,8 @@ static int asf_write_header1(AVFormatContext *s, int64_t file_size,
                     asf->streams[n].stream_language_index = asf->nb_languages;
                     asf->nb_languages++;
                 }
+                if (enc->codec_type == AVMEDIA_TYPE_AUDIO)
+                    audio_language_counts[asf->streams[n].stream_language_index]++;
             }
         } else {
             asf->streams[n].stream_language_index = 128;
@@ -484,6 +487,7 @@ static int asf_write_header1(AVFormatContext *s, int64_t file_size,
     if (asf->nb_languages) {
         int64_t hpos2;
         int i;
+        int nb_audio_languages = 0;
 
         hpos2 = put_header(pb, &ff_asf_language_guid);
         avio_wl16(pb, asf->nb_languages);
@@ -493,6 +497,25 @@ static int asf_write_header1(AVFormatContext *s, int64_t file_size,
         }
         end_header(pb, hpos2);
 
+        for (i = 0; i < asf->nb_languages; i++)
+            if (audio_language_counts[i])
+                nb_audio_languages++;
+
+        if (nb_audio_languages > 1) {
+            hpos2 = put_header(pb, &ff_asf_group_mutual_exclusion_object);
+            ff_put_guid(pb, &ff_asf_mutex_language);
+            avio_wl16(pb, nb_audio_languages);
+            for (i = 0; i < asf->nb_languages; i++) {
+                if (audio_language_counts[i]) {
+                    avio_wl16(pb, audio_language_counts[i]);
+                    for (n = 0; n < s->nb_streams; n++)
+                        if (asf->streams[n].stream_language_index == i && s->streams[n]->codec->codec_type == AVMEDIA_TYPE_AUDIO)
+                            avio_wl16(pb, n + 1);
+                }
+            }
+            end_header(pb, hpos2);
+        }
+
         for (n = 0; n < s->nb_streams; n++) {
             int64_t es_pos;
             if (asf->streams[n].stream_language_index > 127)