Tsearch2 with UTF-8

• Get myspell dictionary from OO site (el_GR.zip) Hmm, this dictionary is just a list of words without ispell tags ! ok, meanwhile will use available ispell dicts. dict.el.utf8.tar.gz is what I used. So, skip convertation to ispell format.
• Convert to ispell format. Copy el_GR.zip myspell to my2ispell subdirectory in tsearch2 distribution.

  gmake ZIPFILE=el_GR LANGUAGE=el_GR
  

• Convert dictionary files to UTF-8.

  iconv -t utf-8 -f iso-8859-7 el_GR.aff > el_GR_utf8.aff
  iconv -t utf-8 -f iso-8859-7 el_GR.dict > el_GR_utf8.dict
  

• Move them to some directory (/tmp/greek.utf8)

  ( mkdir -p /tmp/greek.utf8 && mv el_GR_utf8.aff el_GR_utf8.dict /tmp/greek.utf8 )
  

• Configure postgres and test database (make sure, you have installed PostgreSQL 8.2dev and tsearch2 from distribution).

  initdb -D /usr/local/pgsql-dev/data.el_utf8 --locale=el_GR.utf8
  pg_ctl -D /usr/local/pgsql-dev/data.el_utf8 start
  createdb test
  psql test <  /usr/local/pgsql-dev/share/contrib/tsearch2.sql
  

• Load tsearch2 configuration (tscfg-el.utf8.sql).

  psql test < tscfg-el.utf8.sql
  

Please, notice, tsearch2 configured (tscfg-el.utf8.sql) for testing purposes only !

test=# set client_encoding='ISO-8859-7';
SET
test=# select dict_name,dict_initoption from pg_ts_dict where dict_name = 'el_ispell';
el_ispell | DictFile="tmp/greek.utf8/el_GR_utf8.dic",AffFile="tmp/greek.utf8/el.u8.aff"

test=# select * from pg_ts_cfgmap where ts_name='utf8';
 ts_name |  tok_alias   |  dict_name
 ---------+--------------+-------------
 utf8    | nlhword      | {el_ispell}
 utf8    | nlword       | {el_ispell}
 utf8    | nlpart_hword | {el_ispell}
 (3 rows)

test=# select lexize('el_ispell','ñóñÅ÷øõ‹?õ');
  lexize
-----------
{ñóñÅ÷ø”}
(1 row)
test=# select lexize('el_ispell','?õûõ??ñ‹•?');
    lexize
--------------
{?õûõ??ñ‹ÿ?}
(1 row)