On 01/08/14 13:30, wm4 wrote:
> On Fri,  1 Aug 2014 03:05:43 +0200
> Luca Barbato <[email protected]> wrote:
> 
>> And leverage the video index if the video is just disabled as wm4
>> did in an initial patch.
>> ---
>>  libavformat/flvdec.c | 11 +++++++++--
>>  1 file changed, 9 insertions(+), 2 deletions(-)
>>
>> diff --git a/libavformat/flvdec.c b/libavformat/flvdec.c
>> index 8e4cc5e..d74741d 100644
>> --- a/libavformat/flvdec.c
>> +++ b/libavformat/flvdec.c
>> @@ -57,6 +57,7 @@ typedef struct {
>>      } validate_index[2];
>>      int validate_next;
>>      int validate_count;
>> +    int has_video;
>>  } FLVContext;
>>  
>>  static int flv_probe(AVProbeData *p)
>> @@ -76,11 +77,14 @@ static int flv_probe(AVProbeData *p)
>>  
>>  static AVStream *create_stream(AVFormatContext *s, int codec_type)
>>  {
>> +    FLVContext *flv = s->priv_data;
>>      AVStream *st = avformat_new_stream(s, NULL);
>>      if (!st)
>>          return NULL;
>>      st->codec->codec_type = codec_type;
>>      avpriv_set_pts_info(st, 32, 1, 1000); /* 32 bit pts in ms */
>> +    if (codec_type == AVMEDIA_TYPE_VIDEO)
>> +        flv->has_video = 1;
>>      return st;
>>  }
>>  
>> @@ -818,6 +822,11 @@ skip:
>>              st = create_stream(s, is_audio ? AVMEDIA_TYPE_AUDIO
>>                                             : AVMEDIA_TYPE_VIDEO);
>>          av_dlog(s, "%d %X %d \n", is_audio, flags, st->discard);
>> +
>> +        if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY ||
>> +            !flv->has_video)
>> +            av_add_index_entry(st, pos, dts, size, 0, AVINDEX_KEYFRAME);
>> +
> 
> Why not "if ((flags & FLV_VIDEO_FRAMETYPE_MASK) == FLV_FRAME_KEY || is_audio"
> instead? Maybe it's actually more correct, because it will increase the
> seeking accuracy if video is disabled. Then it could jump to any audio
> packet, instead only to audio packets close to a video keyframe.
> 

I was thinking about it, that would unbreak the samples provided by jb
as well.

lu



_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to