Convert unaccent tests to UTF-8

This makes it easier to add new tests that are specific to Unicode
features.  The files were previously in KOI8-R.

Discussion: https://www.postgresql.org/message-id/8506.1545111362@sss.pgh.pa.us

diff --git a/contrib/unaccent/expected/unaccent.out b/contrib/unaccent/expected/unaccent.out
index b93105e9c7c59bc27c2182609e61206c909e4731..0835e141afbaf5b8e2d747071d4e5cc26df3f707 100644 (file)
--- a/contrib/unaccent/expected/unaccent.out
+++ b/contrib/unaccent/expected/unaccent.out
@@ -6,23 +6,23 @@ SELECT getdatabaseencoding();
  UTF8
 (1 row)
 
-SET client_encoding TO 'KOI8';
+SET client_encoding TO 'UTF8';
 SELECT unaccent('foobar');
  unaccent 
 ----------
  foobar
 (1 row)
 
-SELECT unaccent('£ÌËÁ');
+SELECT unaccent('ёлка');
  unaccent 
 ----------
- ÅÌËÁ
+ елка
 (1 row)
 
-SELECT unaccent('³öéë');
+SELECT unaccent('ЁЖИК');
  unaccent 
 ----------
- åöéë
+ ЕЖИК
 (1 row)
 
 SELECT unaccent('unaccent', 'foobar');
@@ -31,16 +31,16 @@ SELECT unaccent('unaccent', 'foobar');
  foobar
 (1 row)
 
-SELECT unaccent('unaccent', '£ÌËÁ');
+SELECT unaccent('unaccent', 'ёлка');
  unaccent 
 ----------
- ÅÌËÁ
+ елка
 (1 row)
 
-SELECT unaccent('unaccent', '³öéë');
+SELECT unaccent('unaccent', 'ЁЖИК');
  unaccent 
 ----------
- åöéë
+ ЕЖИК
 (1 row)
 
 SELECT ts_lexize('unaccent', 'foobar');
@@ -49,15 +49,15 @@ SELECT ts_lexize('unaccent', 'foobar');
  
 (1 row)
 
-SELECT ts_lexize('unaccent', '£ÌËÁ');
+SELECT ts_lexize('unaccent', 'ёлка');
  ts_lexize 
 -----------
- {ÅÌËÁ}
+ {елка}
 (1 row)
 
-SELECT ts_lexize('unaccent', '³öéë');
+SELECT ts_lexize('unaccent', 'ЁЖИК');
  ts_lexize 
 -----------
- {åöéë}
+ {ЕЖИК}
 (1 row)
 

diff --git a/contrib/unaccent/sql/unaccent.sql b/contrib/unaccent/sql/unaccent.sql
index 310213994f377946288e8967fa52722f8477738e..ba72ab6261c09a9a446da468756942d06159e7cd 100644 (file)
--- a/contrib/unaccent/sql/unaccent.sql
+++ b/contrib/unaccent/sql/unaccent.sql
@@ -3,16 +3,16 @@ CREATE EXTENSION unaccent;
 -- must have a UTF8 database
 SELECT getdatabaseencoding();
 
-SET client_encoding TO 'KOI8';
+SET client_encoding TO 'UTF8';
 
 SELECT unaccent('foobar');
-SELECT unaccent('£ÌËÁ');
-SELECT unaccent('³öéë');
+SELECT unaccent('ёлка');
+SELECT unaccent('ЁЖИК');
 
 SELECT unaccent('unaccent', 'foobar');
-SELECT unaccent('unaccent', '£ÌËÁ');
-SELECT unaccent('unaccent', '³öéë');
+SELECT unaccent('unaccent', 'ёлка');
+SELECT unaccent('unaccent', 'ЁЖИК');
 
 SELECT ts_lexize('unaccent', 'foobar');
-SELECT ts_lexize('unaccent', '£ÌËÁ');
-SELECT ts_lexize('unaccent', '³öéë');
+SELECT ts_lexize('unaccent', 'ёлка');
+SELECT ts_lexize('unaccent', 'ЁЖИК');