On Fri, Dec 06, 2002 at 09:33:10PM +0100, C�dric Dufour (public) wrote:
> Use the 'to_ascii' function to convert your string to the "no accent" ASCII
> equivalent before accomplishing your comparison
>
> SELECT foo FROM table WHERE ( to_ascii(foo) ILIKE to_ascii('caract�res
> accentu�s') );
>
> This does not work with all database locale (LATIN1 is OK, but LATIN9 is
> not).
>
> I was actually wondering if this is efficient enough or if there is any more
> efficient method to accomplish this...
I'd think that something like:
CREATE FUNCTION lower_ascii (text) RETURNS text AS '
BEGIN
RETURN lower(to_ascii($1));
END
' language 'plpgsql';
CREATE INDEX table_lower_ascii ON table(lower_ascii(field));
would perform better, since we can now use this index, whereas we
couldn't do this with ILIKE to_ascii(...).
Also, not sure it's a good idea to use ILIKE simply to get
lower-case-matching. If the user string ends with '%', for instance, it
will match everything-starting-with, which is probably not what the user
meant. Better the check against lower().
There might be a better way specifically-oriented toward
de-accentuation; this is just generic PG advice.
- J.
--
Joel BURTON | [EMAIL PROTECTED] | joelburton.com | aim: wjoelburton
Independent Knowledge Management Consultant
---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?
http://www.postgresql.org/users-lounge/docs/faq.html
