2009/3/18 Sam Mason
<sam@samason.me.uk>
If you can't find anything better in PG; the translate[1] function would
be my best suggestion. Performance should be better than using regular
expressions.
Yeah, that does appear to perform better. I tried the following at it worked for me:
CREATE OR REPLACE FUNCTION unaccent_string(text) RETURNS text AS $$
DECLARE
input_string text := $1;
BEGIN
input_string := translate(input_string, 'âãäåāăąÁÂÃÄÅĀĂĄ', 'aaaaaaaaaaaaaaa');
input_string := translate(input_string, 'èééêëēĕėęěĒĔĖĘĚ', 'eeeeeeeeeeeeeee');
input_string := translate(input_string, 'ìíîïìĩīĭÌÍÎÏÌĨĪĬ', 'iiiiiiiiiiiiiiii');
input_string := translate(input_string, 'óôõöōŏőÒÓÔÕÖŌŎŐ', 'ooooooooooooooo');
input_string := translate(input_string, 'ùúûüũūŭůÙÚÛÜŨŪŬŮ', 'uuuuuuuuuuuuuuuu');
return input_string;
END;
$$ LANGUAGE plpgsql;
Regards
Thom