Bug#22052 Trailing spaces are not removed from UNICODE fields in an index

  Fix: using charset-aware functions cs->cset->lengthsp() and cs->cset->fill()
  instead of single byte code which is not UCS2 compatible.


myisam/mi_key.c:
  Using character-set aware functions to trim and append spaces.
myisam/mi_open.c:
  Initialize charset for BINARY/VARBINARY to my_charset_bin,
      instead of NULL
mysql-test/r/ctype_ucs.result:
  Adding test case
mysql-test/t/ctype_ucs.test:
  Adding test case

myisam/mi_key.c

@@ -42,7 +42,7 @@ static int _mi_put_key_in_record(MI_INFO *info,uint keynr,byte *record);
 uint _mi_make_key(register MI_INFO *info, uint keynr, uchar *key,
 		  const byte *record, my_off_t filepos)
 {
-  byte *pos,*end;
+  byte *pos;
   uchar *start;
   reg1 HA_KEYSEG *keyseg;
   my_bool is_ft= info->s->keyinfo[keynr].flag & HA_FULLTEXT;
@@ -84,18 +84,17 @@ uint _mi_make_key(register MI_INFO *info, uint keynr, uchar *key,
     pos= (byte*) record+keyseg->start;
     if (keyseg->flag & HA_SPACE_PACK)
     {
-      end= pos + length;
       if (type != HA_KEYTYPE_NUM)
       {
-	while (end > pos && end[-1] == ' ')
-	  end--;
+        length= cs->cset->lengthsp(cs, pos, length);
       }
       else
       {
+        byte *end= pos + length;
 	while (pos < end && pos[0] == ' ')
 	  pos++;
+	length=(uint) (end-pos);
       }
-      length=(uint) (end-pos);
       FIX_LENGTH(cs, pos, length, char_length);
       store_key_length_inc(key,char_length);
       memcpy((byte*) key,(byte*) pos,(size_t) char_length);
@@ -358,8 +357,10 @@ static int _mi_put_key_in_record(register MI_INFO *info, uint keynr,
       pos= record+keyseg->start;
       if (keyseg->type != (int) HA_KEYTYPE_NUM)
       {
-	memcpy(pos,key,(size_t) length);
-	bfill(pos+length,keyseg->length-length,' ');
+        memcpy(pos,key,(size_t) length);
+        keyseg->charset->cset->fill(keyseg->charset,
+                                    pos + length, keyseg->length - length,
+                                    ' ');
       }
       else
       {

myisam/mi_open.c

@@ -323,6 +323,8 @@ MI_INFO *mi_open(const char *name, int mode, uint open_flags)
 	      goto err;
 	    }
 	  }
+	  else if (pos->type == HA_KEYTYPE_BINARY)
+	    pos->charset= &my_charset_bin;
 	}
 	if (share->keyinfo[i].flag & HA_SPATIAL)
 	{

mysql-test/r/ctype_ucs.result

@@ -715,6 +715,28 @@ lily
 river
 drop table t1;
 deallocate prepare stmt;
+create table t1 (
+a char(10) unicode not null, 
+index a (a)
+) engine=myisam;
+insert into t1 values (repeat(0x201f, 10));
+insert into t1 values (repeat(0x2020, 10));
+insert into t1 values (repeat(0x2021, 10));
+explain select hex(a) from t1 order by a;
+id	select_type	table	type	possible_keys	key	key_len	ref	rows	Extra
+1	SIMPLE	t1	index	NULL	a	20	NULL	3	Using index
+select hex(a) from t1 order by a;
+hex(a)
+201F201F201F201F201F201F201F201F201F201F
+2020202020202020202020202020202020202020
+2021202120212021202120212021202120212021
+alter table t1 drop index a;
+select hex(a) from t1 order by a;
+hex(a)
+201F201F201F201F201F201F201F201F201F201F
+2020202020202020202020202020202020202020
+2021202120212021202120212021202120212021
+drop table t1;
 CREATE TABLE t1 (id int, s char(5) CHARACTER SET ucs2 COLLATE ucs2_unicode_ci);
 INSERT INTO t1 VALUES (1, 'ZZZZZ'), (1, 'ZZZ'), (2, 'ZZZ'), (2, 'ZZZZZ');
 SELECT id, MIN(s) FROM t1 GROUP BY id;

mysql-test/t/ctype_ucs.test

@@ -452,6 +452,24 @@ select utext from t1 where utext like '%%';
 drop table t1;
 deallocate prepare stmt;
 
+#
+# Bug#22052 Trailing spaces are not removed from UNICODE fields in an index
+#
+create table t1 (
+  a char(10) unicode not null, 
+  index a (a)
+) engine=myisam;
+insert into t1 values (repeat(0x201f, 10));
+insert into t1 values (repeat(0x2020, 10));
+insert into t1 values (repeat(0x2021, 10));
+# make sure "index read" is used
+explain select hex(a) from t1 order by a;
+select hex(a) from t1 order by a;
+alter table t1 drop index a;
+select hex(a) from t1 order by a;
+drop table t1;
+
+
 #
 # Bug #20076: server crashes for a query with GROUP BY if MIN/MAX aggregation
 #             over a 'ucs2' field uses a temporary table