Re: Issue Supporting Emojis in Full Text Search on Ubuntu

On Tue, 30 Jun 2020 20:27:23 -0400
Tom Lane <tgl@sss.pgh.pa.us> wrote:

> With a non-C lc_ctype setting, which characters are considered to be
> letters is entirely up to the locale code supplied by the platform's
> libc. I duplicate your result that U+1F600 is not considered a letter
> by glibc, at least not with lc_ctype = en_US.utf8.  (Perhaps there are
> other locale settings that would consider it a letter?  Not my
> expertise though.)

FWIW, it works ferpectly with a french locale (on Debian buster):

test=# select datname, encoding, datcollate, datctype, datistemplate from
test-# pg_database;
    datname    | encoding | datcollate  |  datctype   | datistemplate
---------------+----------+-------------+-------------+---------------
 postgres      |        6 | fr_FR.utf8  | fr_FR.utf8  | f
 template1     |        6 | fr_FR.utf8  | fr_FR.utf8  | t
 template0     |        6 | fr_FR.utf8  | fr_FR.utf8  | t
 coshonet      |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 ejabberd_test |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 fudforum      |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 postfixadmin  |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 test          |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 testjy        |        6 | fr_FR.UTF-8 | en_US.UTF-8 | f
 zotonic       |        6 | fr_FR.UTF-8 | fr_FR.UTF-8 | f
(10 rows)

test=# create table tst_ctrlu_chars(id varchar(12) primary key, a text);

test=# insert into tst_ctrlu_chars values ('ü', 'aaaaaaaaaaa'),
('ë','bbbbbbbbbb'),('ê','cccccccccc'),('û','dddddddd'),('😀','EEEEEEEE'),('🙉','FFFFFFFFF');
INSERT 0 6

test=# SELECT * FROM tst_ctrlu_chars WHERE id='😀';
 id |    a
----+----------
 😀  | EEEEEEEE
(1 row)


My ¢0.2

Jiff