[MPlayer-cvslog] r24275 - trunk/stream/tvi_vbi.c

[Rich Felker]

On Wed, Aug 29, 2007 at 07:58:01PM +0700, Vladimir Voroshilov wrote:
> 2007/8/29, Rich Felker <dalias at aerifal.cx>:
> > On Tue, Aug 28, 2007 at 08:06:12PM +0200, voroshil wrote:
> > > Author: voroshil
> > > Date: Tue Aug 28 20:06:11 2007
> > > New Revision: 24275
> > >
> > > Log:
> > > Conversion tables for Serbian/Croatian, Ukrainian and Greek charsets.
> >
> > Are all of these just random nonstandard encodings, or could the
> > tables be replaced with calls to iconv?
> >
> 
> All of them are described in ETS 300 706 standard.
> 
> Cyrillic-2 is identical to koi8-r
> Cyrillic-3 may be koi8-u
> 
> I know nothing abouth others.

At first glance the Greek looked like some standard.

> There are only two charset tables remaining (Arabic and Hebrew).

They're probably both the ISO-8859 encodings.

> Most of time conversion (teletext->unicode) is going byte-after-byte
> along with other checks (color, graphics , etc).
> Using iconv in this case will be some overhead, imho.

A generic procedure could just setup the array of actual characters
then call iconv on the whole thing.. Not sure it's worthwhile but I
just cringe every time I see a program duplicating all these tables,
especially because they all eventually end up unmaintained and full of
errors of omissions. Here that might not be a problem if there are a
limited, fixed number of tables though.

Rich