comparison mplayer.c @ 18710:c528c6c518f1

Clean up audio pts handling, make audio pts tracking in the audio-only case work the same as with video.

18709:7ca8f5ab5136

///@}
// Command2Property

int main(int argc,char* argv[]){


char * mem_ptr;

int drop_frame=0;     // current dropping status
int dropped_frames=0; // how many frames dropped since last non-dropped frame
int too_slow_frame_cnt=0;
int too_fast_frame_cnt=0;
// for auto-quality:
float AV_delay=0; // average of A-V timestamp differences
double vdecode_time;
unsigned int lastframeout_ts=0;
/*float time_frame_corr_avg=0;*/ /* unused */

float next_frame_time=0;

} // while(sh_audio)

if(!sh_video) {
  // handle audio-only case:
  if(!quiet) {
    float a_pos = sh_audio->delay - audio_out->get_delay() * playback_speed;
    print_status(a_pos, 0, 0);
  }
  if(d_audio->eof && sh_audio->a_in_buffer_len <= 0 && sh_audio->a_buffer_len <= 0) eof = PT_NEXT_ENTRY;

} else {

//====================== A-V TIMESTAMP CORRECTION: =========================

  current_module="av_sync";

  if(sh_audio){
    double a_pts=0;
    double v_pts=0;

    // unplayed bytes in our and soundcard/dma buffer:
    float delay=playback_speed*audio_out->get_delay()+(float)sh_audio->a_buffer_len/(float)sh_audio->o_bps;

    if (autosync){
      /*
       * If autosync is enabled, the value for delay must be calculated
       * a bit differently.  It is set only to the difference between
       * the audio and video timers.  Any attempt to include the real
       * or corrected delay causes the pts_correction code below to

       * trying to measure.  This keeps the two from competing, but still
       * allows the code to correct for PTS drift *only*.  (Using a delay
       * value here, even a "corrected" one, would be incompatible with
       * autosync mode.)
       */
      delay=sh_audio->delay;
      delay+=(float)sh_audio->a_buffer_len/(float)sh_audio->o_bps;
    }
    delay += sh_audio->a_out_buffer_len*playback_speed/(float)ao_data.bps;

    {
      // PTS = (last timestamp) + (bytes after last timestamp)/(bytes per sec)
      a_pts = sh_audio->pts;
      if (a_pts == MP_NOPTS_VALUE) {
	  // Decoder doesn't support tracking timestamps or demuxer doesn't
	  // set them properly in individual packets, use old inaccurate method
	  a_pts=d_audio->pts;
	  a_pts+=(ds_tell_pts(d_audio)-sh_audio->a_in_buffer_len)/(float)sh_audio->i_bps;
      }
      else
	  a_pts += sh_audio->pts_bytes / (float)sh_audio->o_bps;
    }
    v_pts=sh_video ? sh_video->pts : d_video->pts;

      mp_dbg(MSGT_AVSYNC,MSGL_DBG2,"### A:%8.3f (%8.3f)  V:%8.3f  A-V:%7.4f  \n",a_pts,a_pts-audio_delay-delay,v_pts,(a_pts-delay-audio_delay)-v_pts);

      {
	static int drop_message=0;
        float x;
	AV_delay=(a_pts-delay-audio_delay)-v_pts;
	if(AV_delay>0.5 && drop_frame_cnt>50 && drop_message==0){
	  ++drop_message;
	  mp_msg(MSGT_AVSYNC,MSGL_WARN,MSGTR_SystemTooSlow);
	}
	if (autosync)

          max_pts_correction=default_max_pts_correction;
        else
          max_pts_correction=sh_video->frametime*0.10; // +-10% of time
	if(!frame_time_remaining){ sh_audio->delay+=x; c_total+=x;} // correction
        if(!quiet)
          print_status(a_pts - audio_delay - delay, AV_delay, c_total);
      }
    
  } else {
    // No audio:
    

      float pos = 0;
      if (sh_video)
        pos = sh_video->pts;
      else
      if (sh_audio && audio_out)
        pos = sh_audio->delay - audio_out->get_delay() * playback_speed;
      mp_msg(MSGT_GLOBAL, MSGL_INFO, "ANS_TIME_POSITION=%.1f\n", pos);
    } break;
    case MP_CMD_SWITCH_AUDIO : {
        int current_id = demuxer->audio->id;
        int v = demuxer_switch_audio(demuxer, cmd->args[0].v.i);

18710:c528c6c518f1

///@}
// Command2Property


// Return pts value corresponding to the end point of audio written to the
// ao so far.
static double written_audio_pts(sh_audio_t *sh_audio, demux_stream_t *d_audio)
{
    // first calculate the end pts of audio that has been output by decoder
    double a_pts = sh_audio->pts;
    if (a_pts != MP_NOPTS_VALUE)
	// Good, decoder supports new way of calculating audio pts.
	// sh_audio->pts is the timestamp of the latest input packet with
	// known pts that the decoder has decoded. sh_audio->pts_bytes is
	// the amount of bytes the decoder has written after that timestamp.
	a_pts += sh_audio->pts_bytes / (double) sh_audio->o_bps;
    else {
	// Decoder doesn't support new way of calculating pts (or we're
	// being called before it has decoded anything with known timestamp).
	// Use the old method of audio pts calculation: take the timestamp
	// of last packet with known pts the decoder has read data from,
	// and add amount of bytes read after the beginning of that packet
	// divided by input bps. This will be inaccurate if the input/output
	// ratio is not constant for every audio packet or if it is constant
	// but not accurately known in sh_audio->i_bps.

	a_pts = d_audio->pts;
	// ds_tell_pts returns gets bytes read after last timestamp from
	// demuxing layer, decoder might use sh_audio->a_in_buffer for bytes
	// it has read but not decoded
	a_pts += (ds_tell_pts(d_audio) - sh_audio->a_in_buffer_len) /
	    (double)sh_audio->i_bps;
    }
    // Now a_pts hopefully holds the pts for end of audio from decoder.
    // Substract data in buffers between decoder and audio out.

    // Decoded but not filtered
    a_pts -= sh_audio->a_buffer_len / (double)sh_audio->o_bps;

    // Data that was ready for ao but was buffered because ao didn't fully
    // accept everything to internal buffers yet
    a_pts -= sh_audio->a_out_buffer_len * playback_speed / (double)ao_data.bps;

    return a_pts;
}

// Return pts value corresponding to currently playing audio.
static double playing_audio_pts(sh_audio_t *sh_audio, demux_stream_t *d_audio,
				ao_functions_t *audio_out)
{
    return written_audio_pts(sh_audio, d_audio) - playback_speed *
	audio_out->get_delay();
}


int main(int argc,char* argv[]){


char * mem_ptr;

int drop_frame=0;     // current dropping status
int dropped_frames=0; // how many frames dropped since last non-dropped frame
int too_slow_frame_cnt=0;
int too_fast_frame_cnt=0;
// for auto-quality:
double vdecode_time;
unsigned int lastframeout_ts=0;
/*float time_frame_corr_avg=0;*/ /* unused */

float next_frame_time=0;

} // while(sh_audio)

if(!sh_video) {
  // handle audio-only case:
  if(!quiet) {
    double a_pos = playing_audio_pts(sh_audio, d_audio, audio_out);
    print_status(a_pos, 0, 0);
  }
  if(d_audio->eof && sh_audio->a_in_buffer_len <= 0 && sh_audio->a_buffer_len <= 0) eof = PT_NEXT_ENTRY;

} else {

//====================== A-V TIMESTAMP CORRECTION: =========================

  current_module="av_sync";

  if(sh_audio){
    double a_pts, v_pts;

    if (autosync)
      /*
       * If autosync is enabled, the value for delay must be calculated
       * a bit differently.  It is set only to the difference between
       * the audio and video timers.  Any attempt to include the real
       * or corrected delay causes the pts_correction code below to

       * trying to measure.  This keeps the two from competing, but still
       * allows the code to correct for PTS drift *only*.  (Using a delay
       * value here, even a "corrected" one, would be incompatible with
       * autosync mode.)
       */
	a_pts = written_audio_pts(sh_audio, d_audio) - sh_audio->delay;
    else
	a_pts = playing_audio_pts(sh_audio, d_audio, audio_out);

    v_pts=sh_video ? sh_video->pts : d_video->pts;

      {
	static int drop_message=0;
	double AV_delay = a_pts - audio_delay - v_pts;
	double x;
	if(AV_delay>0.5 && drop_frame_cnt>50 && drop_message==0){
	  ++drop_message;
	  mp_msg(MSGT_AVSYNC,MSGL_WARN,MSGTR_SystemTooSlow);
	}
	if (autosync)

          max_pts_correction=default_max_pts_correction;
        else
          max_pts_correction=sh_video->frametime*0.10; // +-10% of time
	if(!frame_time_remaining){ sh_audio->delay+=x; c_total+=x;} // correction
        if(!quiet)
          print_status(a_pts - audio_delay, AV_delay, c_total);
      }
    
  } else {
    // No audio:
    

      float pos = 0;
      if (sh_video)
        pos = sh_video->pts;
      else
      if (sh_audio && audio_out)
        pos = playing_audio_pts(sh_audio, d_audio, audio_out);
      mp_msg(MSGT_GLOBAL, MSGL_INFO, "ANS_TIME_POSITION=%.1f\n", pos);
    } break;
    case MP_CMD_SWITCH_AUDIO : {
        int current_id = demuxer->audio->id;
        int v = demuxer_switch_audio(demuxer, cmd->args[0].v.i);