On 04/04/2013 08:13 AM, tomas.novosad@linuxbox.cz wrote:
Hello,
i got exactly same problem.
Only the discussed character is different ;-)). When ThunderBird (or ldapsearch, it doesnt matter) send search query to LDAP with some UTF-8 character, the result query to DB (PGSQL in this case) is like (upper(last_name) LIKE '%šEV%')
where the search parameter is: %<lower case utf8 character>EV%
obviously backsql does not correctly handle UTF8 characters.
I can't find any way how to avoid this. If only back-sql would leave the upper case conversion on DB - like this: (upper(last_name) LIKE upper('%šEV%'T
The solution is to augment table ldap_attr_mappings (with non-trivial implications on DN searching and matching) with a field that specifies the encoding for a particular attribute, and convert back and forth any time an operation affects those attributes. Not trivial, but contributions are welcome.
p.