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=c2465153cccb8965b0008186cdb0a94a5...), 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