fix audio detection problem during scan of ffmpeg audio sources

audio frames can be larger than their container packet sizes, but during
scan, we only feed one container packet to the decoder, then reset
the decoder and try the next packet.  so the audio is never detected.

as buffers are tested, they are added to a cache.  the entire cache is
passed to the decoder to scan for info.  the cache is limited to 4KB.


git-svn-id: svn://svn.handbrake.fr/HandBrake/trunk@3195 b64f7644-9d1e-0410-96f1-a4d463321fa5

5 files changed, 99 insertions, 37 deletions

diff --git a/libhb/common.h b/libhb/common.h
index cf1dc02d8..69c6ddea0 100644
--- a/libhb/common.h
+++ b/libhb/common.h
@@ -436,6 +436,7 @@ struct hb_audio_s
 
         hb_esconfig_t config;
         hb_mux_data_t * mux_data;
+        hb_fifo_t     * scan_cache;
     } priv;
 };
 #endif
diff --git a/libhb/decavcodec.c b/libhb/decavcodec.c
index e083d517e..9f3196a85 100644
--- a/libhb/decavcodec.c
+++ b/libhb/decavcodec.c
@@ -389,44 +389,53 @@ static int decavcodecBSInfo( hb_work_object_t *w, const hb_buffer_t *buf,
     uint8_t *buffer = av_malloc( AVCODEC_MAX_AUDIO_FRAME_SIZE );
     int out_size = AVCODEC_MAX_AUDIO_FRAME_SIZE;
     unsigned char *pbuffer;
-    int pos = 0, pbuffer_size;
+    int pos, pbuffer_size;
 
-    while ( pos < buf->size )
+    while ( buf && !ret )
     {
-        int len;
-
-        if (parser != NULL )
-        {
-            len = av_parser_parse2( parser, context, &pbuffer, &pbuffer_size,
-                                    buf->data + pos, buf->size - pos,
-                                    buf->start, buf->start, AV_NOPTS_VALUE );
-        }
-        else
+        pos = 0;
+        while ( pos < buf->size )
         {
-            pbuffer = buf->data;
-            len = pbuffer_size = buf->size;
-        }
-        pos += len;
-        if ( pbuffer_size > 0 )
-        {
-            AVPacket avp;
-            av_init_packet( &avp );
-            avp.data = pbuffer;
-            avp.size = pbuffer_size;
+            int len;
 
-            len = avcodec_decode_audio3( context, (int16_t*)buffer, &out_size, &avp );
-            if ( len > 0 && context->sample_rate > 0 )
+            if (parser != NULL )
             {
-                info->bitrate = context->bit_rate;
-                info->rate = context->sample_rate;
-                info->rate_base = 1;
-                info->channel_layout = 
-                    hb_ff_layout_xlat(context->channel_layout, context->channels);
-                ret = 1;
-                break;
+                len = av_parser_parse2( parser, context, &pbuffer, 
+                                        &pbuffer_size, buf->data + pos, 
+                                        buf->size - pos, buf->start, 
+                                        buf->start, AV_NOPTS_VALUE );
+            }
+            else
+            {
+                pbuffer = buf->data;
+                len = pbuffer_size = buf->size;
+            }
+            pos += len;
+            if ( pbuffer_size > 0 )
+            {
+                AVPacket avp;
+                av_init_packet( &avp );
+                avp.data = pbuffer;
+                avp.size = pbuffer_size;
+
+                len = avcodec_decode_audio3( context, (int16_t*)buffer, 
+                                             &out_size, &avp );
+                if ( len > 0 && context->sample_rate > 0 )
+                {
+                    info->bitrate = context->bit_rate;
+                    info->rate = context->sample_rate;
+                    info->rate_base = 1;
+                    info->channel_layout = 
+                        hb_ff_layout_xlat(context->channel_layout, 
+                                          context->channels);
+                    ret = 1;
+                    break;
+                }
             }
         }
+        buf = buf->next;
     }
+
     av_free( buffer );
     if ( parser != NULL )
         av_parser_close( parser );
diff --git a/libhb/fifo.c b/libhb/fifo.c
index c72b4e10b..3aeeea273 100644
--- a/libhb/fifo.c
+++ b/libhb/fifo.c
@@ -235,6 +235,23 @@ hb_fifo_t * hb_fifo_init( int capacity, int thresh )
     return f;
 }
 
+int hb_fifo_size_bytes( hb_fifo_t * f )
+{
+    int ret = 0;
+    hb_buffer_t * link;
+
+    hb_lock( f->lock );
+    link = f->first;
+    while ( link )
+    {
+        ret += link->size;
+        link = link->next;
+    }
+    hb_unlock( f->lock );
+
+    return ret;
+}
+
 int hb_fifo_size( hb_fifo_t * f )
 {
     int ret;
diff --git a/libhb/internal.h b/libhb/internal.h
index 59753022e..7672c7311 100644
--- a/libhb/internal.h
+++ b/libhb/internal.h
@@ -87,6 +87,7 @@ void          hb_buffer_copy_settings( hb_buffer_t * dst,
 
 hb_fifo_t   * hb_fifo_init( int capacity, int thresh );
 int           hb_fifo_size( hb_fifo_t * );
+int           hb_fifo_size_bytes( hb_fifo_t * );
 int           hb_fifo_is_full( hb_fifo_t * );
 float         hb_fifo_percent_full( hb_fifo_t * f );
 hb_buffer_t * hb_fifo_get( hb_fifo_t * );
diff --git a/libhb/scan.c b/libhb/scan.c
index e14e6712b..1ae5de007 100644
--- a/libhb/scan.c
+++ b/libhb/scan.c
@@ -30,7 +30,7 @@ typedef struct
 
 static void ScanFunc( void * );
 static int  DecodePreviews( hb_scan_t *, hb_title_t * title );
-static void LookForAudio( hb_title_t * title, hb_buffer_t * b );
+static int LookForAudio( hb_title_t * title, hb_buffer_t * b );
 static int  AllAudioOK( hb_title_t * title );
 
 static const char *aspect_to_string( double aspect )
@@ -163,6 +163,11 @@ static void ScanFunc( void * _data )
                 free( audio );
                 continue;
             }
+            if ( audio->priv.scan_cache )
+            {
+                hb_fifo_flush( audio->priv.scan_cache );
+                hb_fifo_close( &audio->priv.scan_cache );
+            }
             j++;
         }
 
@@ -513,9 +518,11 @@ static int DecodePreviews( hb_scan_t * data, hb_title_t * title )
                         vid_buf = NULL;
                     }
                 }
-                else if( ! AllAudioOK( title ) )
+                else 
                 {
-                    LookForAudio( title, buf_es );
+                    if( ! AllAudioOK( title ) )
+                        if ( !LookForAudio( title, buf_es ) )
+                            buf_es = NULL;
                 }
                 if ( buf_es )
                     hb_buffer_close( &buf_es );
@@ -698,6 +705,15 @@ static int DecodePreviews( hb_scan_t * data, hb_title_t * title )
         ++npreviews;
 
 skip_preview:
+        /* Make sure we found audio rates and bitrates */
+        for( j = 0; j < hb_list_count( title->list_audio ); j++ )
+        {
+            hb_audio_t * audio = hb_list_item( title->list_audio, j );
+            if ( audio->priv.scan_cache )
+            {
+                hb_fifo_flush( audio->priv.scan_cache );
+            }
+        }
         if ( vid_buf )
             hb_buffer_close( &vid_buf );
     }
@@ -809,7 +825,7 @@ skip_preview:
  * aren't (e.g., some European DVD Teletext streams use the same IDs as US ATSC
  * AC-3 audio).
  */
-static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
+static int LookForAudio( hb_title_t * title, hb_buffer_t * b )
 {
     int i;
 
@@ -830,9 +846,20 @@ static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
     if( !audio || audio->config.in.bitrate != 0 )
     {
         /* not found or already done */
-        return;
+        return 1;
     }
 
+    if ( audio->priv.scan_cache == NULL )
+        audio->priv.scan_cache = hb_fifo_init( 16, 16 );
+
+    if ( hb_fifo_size_bytes( audio->priv.scan_cache ) >= 4096 )
+    {
+        hb_buffer_t * tmp;
+        tmp = hb_fifo_get( audio->priv.scan_cache );
+        hb_buffer_close( &tmp );
+    }
+    hb_fifo_push( audio->priv.scan_cache, b );
+
     hb_work_object_t *w = hb_codec_decoder( audio->config.in.codec );
 
     if ( w == NULL || w->bsinfo == NULL )
@@ -845,6 +872,7 @@ static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
     hb_work_info_t info;
     w->audio = audio;
     w->codec_param = audio->config.in.codec_param;
+    b = hb_fifo_see( audio->priv.scan_cache );
     int ret = w->bsinfo( w, b, &info );
     if ( ret < 0 )
     {
@@ -856,8 +884,11 @@ static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
     if ( !info.bitrate )
     {
         /* didn't find any info */
-        return;
+        return 0;
     }
+    hb_fifo_flush( audio->priv.scan_cache );
+    hb_fifo_close( &audio->priv.scan_cache );
+
     audio->config.in.samplerate = info.rate;
     audio->config.in.bitrate = info.bitrate;
     audio->config.in.channel_layout = info.channel_layout;
@@ -886,7 +917,7 @@ static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
             audio->config.lang.description );
  
     free( w );
-    return;
+    return 1;
 
     // We get here if there's no hope of finding info on an audio bitstream,
     // either because we don't have a decoder (or a decoder with a bitstream
@@ -898,7 +929,10 @@ static void LookForAudio( hb_title_t * title, hb_buffer_t * b )
     if ( w )
         free( w );
 
+    hb_fifo_flush( audio->priv.scan_cache );
+    hb_fifo_close( &audio->priv.scan_cache );
     hb_list_rem( title->list_audio, audio );
+    return -1;
 }
 
 /*