[MPlayer-dev-eng] [PATCH] mpeg pass-through for hardware audio decoders

[Nico Sabbi]

Reimar Döffinger wrote:

>Hi,
>On Sat, Jan 14, 2006 at 11:11:42PM +0100, Nico Sabbi wrote:
>  
>
>>+static ad_info_t info = 
>>+{
>>+	"MPEG audio pass-through (fake decoder)",
>>+	"hwmpa",
>>+	"Nico",
>>+	"???",
>>+	""
>>+};
>>    
>>
>
>I admit I never looked it up myself before, but this "says" that you are
>the maintainer and "???" is the author?
>
>  
>
>>+	while(1)
>>+	{
>>+		len = demux_read_data(sh->ds,&sh->a_in_buffer[sh->a_in_buffer_len],sh->a_in_buffer_size-sh->a_in_buffer_len);
>>+		if(len > 0)
>>+			sh->a_in_buffer_len += len;
>>+		while(cnt + 4 < sh->a_in_buffer_len)
>>+		{
>>+			x = mp_get_mp3_header(&(sh->a_in_buffer[cnt]), chans, srate, spf, mpa_layer, br);
>>+			if(x > 0)
>>+			{
>>+				*n = x;
>>+				return cnt;
>>+			}
>>+			cnt++;
>>+		}
>>+		if(len <= 0)
>>+			break;
>>+	}
>>    
>>
>
>well, a do { ... } while (len > 0); would make more sense IMHO.
>  
>

indeed, changed

>But maybe you could do something like the demux_pattern_3 function in my
>es parsing patch 
>

I added something similar; feel free to optimize it with something better

>(I assume you checked that ds_get_packet doesn't
>already return the data correctly frame-by-frame)?.
>  
>
mpeg-[pt]s surely doesn't, and other formats don't do it either

>E.g. introduce a demux_pattern_mask that allows an arbitrary bitmask instead
>of only complete 3-byte patterns. I guess speed is not so much of an issue here.
>It might also make the skip-one-frame thing easier.
>  
>

simplified with the addition of the count of frames to search.
Thanks for your advices.
    Nico