[Koha-devel] ICU chains

[LAURENT Henri-Damien]

Le 30/11/2010 18:09, Ian Walls a écrit :
> Fellow Kohackers,
> 
> 
> There has been some talk before of using ICU chains with Zebra in order
> to resolve issues with diacritic search.  I've looked at the patch from
> BibLibre
> (http://git.biblibre.com/?p=koha;a=commit;h=c2465153cccb8965b0008186cdb0a94a57317849),
> and it seems pretty straightfoward, but it's nearly a year old, and I'm
> wondering if any work has been done since then.  Does this commit
> resolve the issue (at least for French diacritics)?
> 
> There has also been some recent posting on handling both Roman and
> Devangari script
> (http://old.nabble.com/Re:-display-search-result-problem-p28106148.html).  It
> references the BibLibre commit.
> 
> If this methodology is working, what should we do to get it committed to
> Koha?  I'd imagine that we'd need to generalize it to work with all
> languages... either by making a master set of rules, or having different
> xml files that can be called depending on your language choice... ideas?
>
> Cheers,
> 
> 
> -Ian
> 
Ian,

This methodology and icuchain is sure working for Korean, Hebrew,
Georgian, Russian. It might be different for Arabic languages, since in
that language, diacritics are meaningfull. But we had scarce return for
those languages.
But all in all, the commit said, and the icu chain modification
published on 3.4/BibLibre-various (which is fixing a problem on multiple
word searching) will suit you.
Hope that helps.
-- 
Henri-Damien LAURENT