[FFmpeg-devel] [PATCH] encode mov metadata to UTF-8

Reimar Döffinger Reimar.Doeffinger
Tue Feb 2 20:43:27 CET 2010


On Tue, Feb 02, 2010 at 08:24:39PM +0100, Michael Niedermayer wrote:
> On Tue, Feb 02, 2010 at 07:45:45PM +0100, Reimar D?ffinger wrote:
> > Hello,
> > not sure if it is correct, since it assumes metadata is store in latin-1 for
> > all the cases that code handles, however it fixes the output for
> > http://samples.mplayerhq.hu/V-codecs/SVQ3/Vertical400kbit.sorenson3.mov
> > correctly displaying the (c) sign (on and UTF-8 console at least).
> 
> what about some function that takes a string and returns a guess if its
> more likely latin1 or utf8 ?
> I think the probability of random latin1 to parse as valid utf8 is low

Well, we may know if it's UTF-8 like that, but we still don't know what it
is otherwise (which is the problem with real).
I haven't bothered to understand the code, but I think mov has some "legacy"
metadata format that actually has to be in latin1, in which case this
code might be completely correct without any such function.



More information about the ffmpeg-devel mailing list