sql: unaccent function

[verglor]

FEATURE REQUEST

Please implement PostgreSQL's unaccent() function.

[knz]

Note: This feature is conditional on supporting PostgreSQL's full text search features, a much larger project than just a SQL built-in function.

See #7821.

[bdarnell]

As a workaround for some use cases, the magic undocumented collations (cockroachdb/docs#1471) can be used instead. A "level1" collation performs accent-insensitive comparisons so a unique constraint on a column of this type will perform accent-insensitive duplicate checks, for example:

[email protected]:58318/defaultdb> create table t (s string collate en_u_ks_level1 primary key);
CREATE TABLE

Time: 4.625ms

[email protected]:58318/defaultdb> insert into t values ('hotel'::string collate en_u_ks_level1);
INSERT 1

Time: 3.332ms

[email protected]:58318/defaultdb> insert into t values ('hôtel'::string collate en_u_ks_level1);
pq: duplicate key value (s)=('hotel' COLLATE en_u_ks_level1) violates unique constraint "primary"

[jordanlewis]

Closed by #54628. Thanks @mknycha!