[FFmpeg-soc] [FFmpeg-devel] [Patch]GSoC 2008 qualification task TS Muxer

[zhentan feng]

2008/3/28, Baptiste Coudurier <baptiste.coudurier at smartjog.com>:
> Hi,
>
>
>  On Thu, Mar 27, 2008 at 10:52:09PM +0800, zhentan feng wrote:
>  > Index: mpegenc.c
>  > ===================================================================
>
> > --- mpegenc.c (revision 2048)
>  > +++ mpegenc.c (working copy)
>  > @@ -266,11 +266,13 @@
>  >  static int mpeg_mux_init(AVFormatContext *ctx)
>  >  {
>
> >      MpegMuxContext *s = ctx->priv_data;
>
> > -    int bitrate, i, mpa_id, mpv_id, mps_id, ac3_id, dts_id, lpcm_id, j;
>  > +    int bitrate, i;
>  >      AVStream *st;
>  >      PESStream *stream;
>  >      int audio_bitrate;
>  >      int video_bitrate;
>  > +    int *ps_audio_bound = &(s->audio_bound);
>  > +    int *ps_video_bound = &(s->video_bound);
>  >
>  > [...]
>
> >
>  > +    if(ff_pes_muxer_init(ctx,ps_audio_bound,ps_video_bound) != 0)
>  >
>
>
> Code seems to only count audio and video streams, I don't think
>  ff_pes_muxer_init needs to be extented, only count them in
>  mpeg_mux_init (a small for loop should do the trick).
>
yes.
ps_audio_bound, and ps_video_bound are just flags to keep off TS can
not execute some codes when call the function ff_pes_muxer_init().
I review the codes again , find  mepg_mux_init() in mpegenc.c and
mpegts_write_header() in mpegtsenc.c actually have small common codes.

Moreover, mpegtsenc.c initial the stream data as below:
for(i=0;i<ctx->nb_streams;i++) {
        st = ctx->streams[i];
        stream = st->priv_data;

but mpegenc.c initial the stream like this:
 for(i=0;i<ctx->nb_streams;i++) {
        st = ctx->streams[i];
        stream = av_mallocz(sizeof(StreamInfo));
        if (!stream)
            goto fail;
        st->priv_data = stream;

Thus, I think it is stiff resued without flags to sign PS or TS stream.
So, I leave it just as svn-soc original.
>  >
>  > [...]
>  >
>  >  /**
>  > + * Caculate the PES header to flush
>  > + * @param[in] ps_flag    the sign for PS, '1' is PS, '0' is TS
>  > + * @param[in] is_mpeg2 the pointer point to PS struct
>  > + * @param[in] is_dvd   the pointer point to PS struct
>  >
>  > [...]
>  >
>  > + */
>  > +void ff_pes_cal_header(int ps_flag,int *is_mpeg2,int *is_dvd,int id,PESStream *stream,
>  > +          int *packet_size,int *header_len,int64_t *pts,int64_t *dts,
>  > +          int *payload_size,int *startcode,int *stuffing_size,
>  > +          int *trailer_size,int *pad_packet_bytes);
>  > +
>  > +/**
>  >
>
>  I think some flags should be used for different variants:
>
>  #define FMT_MPEG2 0x01
>  #define FMT_VCD   0x02 | FMT_MPEG2
>  #define FMT_SVCD  0x04 | FMT_MPEG2
>  #define FMT_DVD   0x08 | FMT_MPEG2
>  #define FMT_TS    0x10 | FMT_MPEG2
>
>  then add a "format" field in PESStream.
>
>  You will be able to check with (s->format & FMT_MPEG2).
>
>  Beware of mpeg1system muxer though.
>
>  It should be simpler and cleaner, and will avoid passing 3 args.
>
>  This needs a separate patch.

The patch names "ff_pes_cal_header_3-29.patch" show the changes.


>
>  [...]
>
>
>  --
>  Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
>  SMARTJOG SAS                                     http://www.smartjog.com
>  Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
>  Phone: +33 1 49966312
>
> _______________________________________________
>  FFmpeg-soc mailing list
>  FFmpeg-soc at mplayerhq.hu
>  https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-soc
>


-- 
Best wishes~
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ff_pes_cal_header_3-29.patch
Type: application/octet-stream
Size: 12307 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/attachments/20080330/3c1d5f35/attachment.obj>