From 600b5fea57b32ef60f32bec61a12f7290c42bae1 Mon Sep 17 00:00:00 2001
From: unknown <bar@mysql.com>
Date: Thu, 1 Sep 2005 19:31:08 +0500
Subject: [PATCH] Bug#12891:   UNION doesn't return DISTINCT result for
 multi-byte characters hp_hash.c:   This piece of code was pretty wrong, looks
   like no necessary changes were made after   cut-and-paste from fixed length
 segment processing.   Itroduced two new variables safe_length1 and  
 safe_length2 to remember the original lengths.   Fixing my_charpos and
 set_if_smaller calls   to pass correct parameters. ctype_utf8.result,
 ctype_utf8.test:   adding test case

heap/hp_hash.c:
   Bug#12891:
  UNION doesn't return DISTINCT result for multi-byte characters
  This piece of code was pretty wrong.
mysql-test/t/ctype_utf8.test:
  adding test case
mysql-test/r/ctype_utf8.result:
  adding test case
---
 heap/hp_hash.c                 | 10 ++++++----
 mysql-test/r/ctype_utf8.result | 12 ++++++++++++
 mysql-test/t/ctype_utf8.test   | 11 +++++++++++
 3 files changed, 29 insertions(+), 4 deletions(-)

diff --git a/heap/hp_hash.c b/heap/hp_hash.c
index d643f776731..77f3cf6d80b 100644
--- a/heap/hp_hash.c
+++ b/heap/hp_hash.c
@@ -551,11 +551,13 @@ int hp_rec_key_cmp(HP_KEYDEF *keydef, const byte *rec1, const byte *rec2,
       }
       if (cs->mbmaxlen > 1)
       {
+        uint safe_length1= char_length1;
+        uint safe_length2= char_length2;
         uint char_length= seg->length / cs->mbmaxlen;
-        char_length1= my_charpos(cs, pos1, pos1 + char_length1, char_length1);
-        set_if_smaller(char_length1, seg->length);
-        char_length2= my_charpos(cs, pos2, pos2 + char_length2, char_length2);
-        set_if_smaller(char_length2, seg->length);
+        char_length1= my_charpos(cs, pos1, pos1 + char_length1, char_length);
+        set_if_smaller(char_length1, safe_length1);
+        char_length2= my_charpos(cs, pos2, pos2 + char_length2, char_length);
+        set_if_smaller(char_length2, safe_length2);
       }
 
       if (cs->coll->strnncollsp(seg->charset,
diff --git a/mysql-test/r/ctype_utf8.result b/mysql-test/r/ctype_utf8.result
index 64c693a292a..ee109887dd3 100644
--- a/mysql-test/r/ctype_utf8.result
+++ b/mysql-test/r/ctype_utf8.result
@@ -1023,3 +1023,15 @@ aa
 xxx
 yyy
 DROP TABLE t1;
+set names utf8;
+create table t1 (a char(1)) default character set utf8;
+create table t2 (a char(1)) default character set utf8;
+insert into t1 values('a'),('a'),(0xE38182),(0xE38182);
+insert into t1 values('i'),('i'),(0xE38184),(0xE38184);
+select * from t1 union distinct select * from t2;
+a
+a
+あ
+i
+い
+drop table t1,t2;
diff --git a/mysql-test/t/ctype_utf8.test b/mysql-test/t/ctype_utf8.test
index ce259f465d9..eda64b82f38 100644
--- a/mysql-test/t/ctype_utf8.test
+++ b/mysql-test/t/ctype_utf8.test
@@ -857,3 +857,14 @@ SELECT DISTINCT id FROM t1;
 SELECT DISTINCT id FROM t1 ORDER BY id;
 
 DROP TABLE t1;
+
+#
+# Bug#12891: UNION doesn't return DISTINCT result for multi-byte characters
+#
+set names utf8;
+create table t1 (a char(1)) default character set utf8;
+create table t2 (a char(1)) default character set utf8;
+insert into t1 values('a'),('a'),(0xE38182),(0xE38182);
+insert into t1 values('i'),('i'),(0xE38184),(0xE38184);
+select * from t1 union distinct select * from t2;
+drop table t1,t2;