Re: some progress with the spell checker

On Friday 24 June 2005 00:29, Heiko Evermann wrote:
> Well,
>
> my problems might have to do with character coding. But aspell via the
> command line
>
> > aspell -a -S -d <lang code>
>
> is doing fine. I get an interactive program, I can enter a text and get
> a result.
>

> Could you do me a favour: could some of you (it does not take long)
> enter a text somewhere in a po-file, a text containing special characters.
> I think that any language of yours should contain some, e.g. if it is a
> french word that you happen to use. Do some nonsense with the word to make
> sure that it is misspelled and tell me, whether it is marked. And if it
> works, please tell me
> * which text did you use?
> * is it aspell or ispell
> * which language code

I have tested with a German desktop_kdebase.po (I do not know which old 
version it is exactly however).

With ispell, old German dictionary, ISO-8859-1 in KBabel trunk

>
> I would really like to be able to limit the scope of the error. My fear is,
> that any word with special characters (like french accents) will not be
> checked at all. That would be highly undesirable.

For me, German umlauts seem to work.

>
> I would like to know about the scope of the problem. Because perhaps the
> only viable solution would be to port KBabel to KSpell2 (like KOffice did
> some time ago). And I would like to know, whether or not that is
> neccessary.

As Stanislav Visnovsky has written, porting to KSpell2 will be necessary for 
KDE4. As anyway many major modifications seem to be needed for Qt4/KDE4, it 
would be nice if that problem could be postponed until then too.

Have a nice day!

>
> Kind regards,
>
> Heiko Evermann
> (who really needs umlaut checking for Low Saxon)
>
> > --- Ursprüngliche Nachricht ---
> > Von: Lars Ivar Igesund <[email protected]>
> > An: [email protected]
> > Betreff: Re: [Kbabel] some progress with the spell checker
> > Datum: Thu, 23 Jun 2005 20:30:35 +0100
> >
> > Heiko Evermann wrote:
> > >Hi Nicolas,
> > >
> > >>>For the umlauts, I already have an idea: the regexps will not treat it
> > >>
> > >>as a
> > >>
> > >>>letter. I hope that the character classes of QT will solve that.
> > >>
> > >>Otherwise you could try to use . for any character or \S for any
> > >>non-space.
> > >
> > >The problem is a bit harder that I thought. I have added a debug message
> > >about the words that are treated in KBabelHighlighter::isMisspelled.
> > >You should get them, too, if you compile fresh from svn. The line is
> >
> > split
> >
> > >correctly into words. The word with umlaut is then handed over to
> > >
> > >>mSpell->checkWord(word, false);
> > >
> > >Any ideas?
> > >
> > >Kind regards,
> > >
> > >Heiko Evermann
> >
> > If I remember correctly, characters with umlaut use two bytes in a utf8
> > string, and this can in some cases mean that no match is found, even in
> > the same encoding. If I remember wrong, it might still be an encoding
> > problem, in which case I wish you luck :)
> >
> > Regards,
> > Lars Ivar Igesund
> >
> > _______________________________________________
> > kbabel mailing list
> > [email protected]
> > https://mail.kde.org/mailman/listinfo/kbabel