Re: utf-8 & ncurses

[Thomas Dickey]

On Tue, 17 Aug 2010, folkert wrote:

> > > I'm trying to get ncurses and utf-8 to work with each other.
> > > Let's say I have a string 'string' and an index 'loop'. Now in this
> > > string some utf-8 characters may exist. I try to filter these out and
> > > print them.
> > >
> > > #include <ncursesw/ncurses.h>
> > >
> > >         if ((string[loop] & 0xe0) == 0xc0)
> > >         {
> > >                 wadd_wch(win -> win, (string[loop] << 8) + string[loop + 1]);
> > >                 loop += 1;
> > >         }
> > >         else if ((string[loop] & 0xf0) == 0xe0)
> > >         {
> > >                 wadd_wch(win -> win, (string[loop] << 16) + (string[loop + 1] 
> > > << 8) + string[loop + 2]);
> > >                 loop += 2;
> > >         }
> > > ....
> > >
> > > This doesn't work. No special characters are shown.
> >
> > The conversion doesn't look right.  Basically, UTF-8 uses a few bits from
> > each byte to tell where it is in a multibyte sequence, and a few bits from
> > each byte are masked into the result.
>
> So what you mean is that I need to strip of those markerbits,
> concatenate the remaining bits into some variable and then use waddch to
> print it?

no - I was saying that ncurses' waddch can print the UTF-8 directly,
assuming that setlocale has already told it that it's using UTF-8 
encoding.

If the encoding doesn't match the locale, it couldn't work anyway,
except for the special case where the locale is using ISO-8859-1.
In that special case, it's possible to do your own explicit conversion
and use ncurses to print the codes.

--
Thomas E. Dickey
http://invisible-island.net
ftp://invisible-island.net