core: allow disabling display of "album art" in audio files

ffmpeg pretends that image attachments (such as contained in ID3v2
metadata) are video streams. It injects the attached pictures as packets
into the packet stream received with av_read_frame().

Add the --audio-display option to allow configuring whether attached
pictures should be displayed. The default behavior doesn't change
(images are displayed).

Identify video streams, that are actually image attachments, with "[P]"
in the terminal output.

Modify the default stream selection such that real video streams are
preferred over attached pictures. (This is just for robustness; I do not
know of any samples where images are added before actual video streams
and could lead to bad default stream selection with the old code.)

8 files changed, 32 insertions, 6 deletions

diff --git a/DOCS/man/en/options.rst b/DOCS/man/en/options.rst
index 253bea472b..1d8798f260 100644
--- a/DOCS/man/en/options.rst
+++ b/DOCS/man/en/options.rst
@@ -186,6 +186,16 @@
     name to force it, this will skip some checks! Give the demuxer name as
     printed by ``--audio-demuxer=help``. ``--audio-demuxer=audio`` forces MP3.
 
+--audio-display=<no|attachment>
+    Setting this option to ``attachment`` (default) will display image
+    attachments when playing audio files. It will display the first image
+    found, and additional images are available as video streams.
+
+    Setting this option to ``no`` disables display of video entirely when
+    playing audio files.
+
+    This option has no influence on files with normal video tracks.
+
 --audiofile=<filename>
     Play audio from an external file (WAV, MP3 or Ogg Vorbis) while viewing a
     movie.
diff --git a/core/cfg-mplayer.h b/core/cfg-mplayer.h
index 6534672223..762c13a2bb 100644
--- a/core/cfg-mplayer.h
+++ b/core/cfg-mplayer.h
@@ -394,6 +394,9 @@ const m_option_t common_opts[] = {
     OPT_STRINGLIST("alang", audio_lang, 0),
     OPT_STRINGLIST("slang", sub_lang, 0),
 
+    OPT_CHOICE("audio-display", audio_display, 0,
+               ({"no", 0}, {"attachment", 1})),
+
     OPT_STRING("quvi-format", quvi_format, 0),
 
     { "rawaudio", (void *)&demux_rawaudio_opts, CONF_TYPE_SUBCONFIG, 0, 0, 0, NULL},
diff --git a/core/defaultopts.c b/core/defaultopts.c
index 9f746b3dc1..4338e10b0a 100644
--- a/core/defaultopts.c
+++ b/core/defaultopts.c
@@ -44,6 +44,7 @@ void set_default_mplayer_options(struct MPOpts *opts)
         .audio_id = -1,
         .video_id = -1,
         .sub_id = -1,
+        .audio_display = 1,
         .sub_visibility = 1,
         .extension_parsing = 1,
         .audio_output_channels = 2,
diff --git a/core/mp_core.h b/core/mp_core.h
index 92803d2dd0..5a7926dee9 100644
--- a/core/mp_core.h
+++ b/core/mp_core.h
@@ -89,6 +89,7 @@ struct track {
 
     char *title;
     bool default_track;
+    bool attached_picture;
     char *lang;
 
     // If this track is from an external file (e.g. subtitle file).
diff --git a/core/mplayer.c b/core/mplayer.c
index 99fd4b46ed..5dc35479ba 100644
--- a/core/mplayer.c
+++ b/core/mplayer.c
@@ -320,6 +320,8 @@ static void print_stream(struct MPContext *mpctx, struct track *t, int id)
         mp_msg(MSGT_CPLAYER, MSGL_INFO, " --%s=%s", langopt, t->lang);
     if (t->default_track)
         mp_msg(MSGT_CPLAYER, MSGL_INFO, " (*)");
+    if (t->attached_picture)
+        mp_msg(MSGT_CPLAYER, MSGL_INFO, " [P]");
     if (t->title)
         mp_msg(MSGT_CPLAYER, MSGL_INFO, " '%s'", t->title);
     mp_msg(MSGT_CPLAYER, MSGL_INFO, " (");
@@ -914,6 +916,7 @@ static struct track *add_stream_track(struct MPContext *mpctx,
         .demuxer_id = stream->demuxer_id,
         .title = stream->title,
         .default_track = stream->default_track,
+        .attached_picture = stream->attached_picture,
         .lang = stream->common_header->lang,
         .under_timeline = under_timeline,
         .demuxer = stream->demuxer,
@@ -3592,14 +3595,16 @@ static bool compare_track(struct track *t1, struct track *t2, char **langs)
         return l1 > l2;
     if (t1->default_track != t2->default_track)
         return t1->default_track;
+    if (t1->attached_picture != t2->attached_picture)
+        return !t1->attached_picture;
     return t1->user_tid <= t2->user_tid;
 }
 static struct track *select_track(struct MPContext *mpctx,
-                                  enum stream_type type, int tid, char **langs,
-                                  bool select_fallback)
+                                  enum stream_type type, int tid, char **langs)
 {
     if (tid == -2)
         return NULL;
+    bool select_fallback = type == STREAM_VIDEO || type == STREAM_AUDIO;
     struct track *pick = NULL;
     for (int n = 0; n < mpctx->num_tracks; n++) {
         struct track *track = mpctx->tracks[n];
@@ -3613,6 +3618,8 @@ static struct track *select_track(struct MPContext *mpctx,
     if (pick && !select_fallback && !pick->is_external
         && !match_lang(langs, pick->lang) && !pick->default_track)
         pick = NULL;
+    if (pick && pick->attached_picture && !mpctx->opts.audio_display)
+        pick = NULL;
     return pick;
 }
 
@@ -3990,13 +3997,13 @@ goto_enable_cache: ;
     check_previous_track_selection(mpctx);
 
     mpctx->current_track[STREAM_VIDEO] =
-        select_track(mpctx, STREAM_VIDEO, mpctx->opts.video_id, NULL, true);
+        select_track(mpctx, STREAM_VIDEO, mpctx->opts.video_id, NULL);
     mpctx->current_track[STREAM_AUDIO] =
         select_track(mpctx, STREAM_AUDIO, mpctx->opts.audio_id,
-                     mpctx->opts.audio_lang, true);
+                     mpctx->opts.audio_lang);
     mpctx->current_track[STREAM_SUB] =
         select_track(mpctx, STREAM_SUB, mpctx->opts.sub_id,
-                     mpctx->opts.sub_lang, false);
+                     mpctx->opts.sub_lang);
 
     demux_info_print(mpctx->master_demuxer);
     print_file_properties(mpctx, mpctx->filename);
diff --git a/core/options.h b/core/options.h
index f6e2a169f0..da4449cb11 100644
--- a/core/options.h
+++ b/core/options.h
@@ -85,6 +85,7 @@ typedef struct MPOpts {
     int sub_id;
     char **audio_lang;
     char **sub_lang;
+    int audio_display;
     int sub_visibility;
     char *quvi_format;
 
diff --git a/demux/demux_lavf.c b/demux/demux_lavf.c
index b3ae1d6315..7c73ac6085 100644
--- a/demux/demux_lavf.c
+++ b/demux/demux_lavf.c
@@ -438,6 +438,8 @@ static void handle_stream(demuxer_t *demuxer, AVFormatContext *avfc, int i)
         priv->vstreams[priv->video_streams] = i;
         sh_video->libav_codec_id = codec->codec_id;
         sh_video->gsh->lavf_codec_tag = lavf_codec_tag;
+        if (st->disposition & AV_DISPOSITION_ATTACHED_PIC)
+            sh_video->gsh->attached_picture = true;
         bih = calloc(sizeof(*bih) + codec->extradata_size, 1);
 
         if (codec->codec_id == CODEC_ID_RAWVIDEO) {
diff --git a/demux/stheader.h b/demux/stheader.h
index b5f11c0ecc..efcec02d7b 100644
--- a/demux/stheader.h
+++ b/demux/stheader.h
@@ -58,7 +58,8 @@ struct sh_stream {
     int lavf_codec_tag;
 
     char *title;
-    bool default_track;
+    bool default_track;         // container default track flag
+    bool attached_picture;      // stream is a picture (such as album art)
 
     // shouldn't exist type of stuff
     struct MPOpts *opts;