BDB: Bug#4531: unique key prefix interacts poorly with utf8,

Bug#4594 column index make = failed for gbk


myisam/mi_key.c:
  cleanup

myisam/mi_key.c

@@ -25,9 +25,12 @@
 
 #define CHECK_KEYS
 
-#define FIX_LENGTH if (length > char_length) \
-                     char_length= my_charpos(cs, pos, pos+length, char_length); \
-                   set_if_smaller(char_length,length); \
+#define FIX_LENGTH                                                          \
+            do {                                                            \
+              if (length > char_length)                                     \
+                char_length= my_charpos(cs, pos, pos+length, char_length);  \
+              set_if_smaller(char_length,length);                           \
+            } while(0)
 
 static int _mi_put_key_in_record(MI_INFO *info,uint keynr,byte *record);
 
@@ -157,7 +160,7 @@ uint _mi_make_key(register MI_INFO *info, uint keynr, uchar *key,
     FIX_LENGTH;
     memcpy((byte*) key, pos, char_length);
     if (length > char_length)
-      bfill(key+char_length, length-char_length, ' ');
+      cs->cset->fill(cs, key+char_length, length-char_length, ' ');
     key+= length;
   }
   _mi_dpointer(info,key,filepos);
@@ -267,7 +270,7 @@ uint _mi_pack_key(register MI_INFO *info, uint keynr, uchar *key, uchar *old,
     FIX_LENGTH;
     memcpy((byte*) key, pos, char_length);
     if (length > char_length)
-      bfill(key+char_length, length-char_length, ' ');
+      cs->cset->fill(cs,key+char_length, length-char_length, ' ');
     key+= length;
     k_length-=length;
   }

mysql-test/r/ctype_utf8.result

@@ -412,6 +412,36 @@ select c as c_a from t1 where c='б';
 c_a
 б
 drop table t1;
+create table t1 (
+c char(10) character set utf8,
+unique key a (c(1))
+) engine=bdb;
+insert into t1 values ('a'),('b'),('c'),('d'),('e'),('f');
+insert into t1 values ('aa');
+ERROR 23000: Duplicate entry 'aa' for key 1
+insert into t1 values ('aaa');
+ERROR 23000: Duplicate entry 'aaa' for key 1
+insert into t1 values ('б');
+insert into t1 values ('бб');
+ERROR 23000: Duplicate entry 'б' for key 1
+insert into t1 values ('ббб');
+ERROR 23000: Duplicate entry 'б' for key 1
+select c as c_all from t1 order by c;
+c_all
+a
+b
+c
+d
+e
+f
+б
+select c as c_a from t1 where c='a';
+c_a
+a
+select c as c_a from t1 where c='б';
+c_a
+б
+drop table t1;
 create table t1 (c varchar(30) character set utf8 collate utf8_bin, unique(c(10)));
 insert into t1 values ('1'),('2'),('3'),('x'),('y'),('z');
 insert into t1 values ('aaaaaaaaaa');
@@ -541,6 +571,36 @@ c_a
 б
 drop table t1;
 create table t1 (
+c char(10) character set utf8 collate utf8_bin,
+unique key a (c(1))
+) engine=bdb;
+insert into t1 values ('a'),('b'),('c'),('d'),('e'),('f');
+insert into t1 values ('aa');
+ERROR 23000: Duplicate entry 'aa' for key 1
+insert into t1 values ('aaa');
+ERROR 23000: Duplicate entry 'aaa' for key 1
+insert into t1 values ('б');
+insert into t1 values ('бб');
+ERROR 23000: Duplicate entry 'б' for key 1
+insert into t1 values ('ббб');
+ERROR 23000: Duplicate entry 'б' for key 1
+select c as c_all from t1 order by c;
+c_all
+a
+b
+c
+d
+e
+f
+б
+select c as c_a from t1 where c='a';
+c_a
+a
+select c as c_a from t1 where c='б';
+c_a
+б
+drop table t1;
+create table t1 (
 str varchar(255) character set utf8 not null,
 key str  (str(2))
 ) engine=myisam;
@@ -570,3 +630,13 @@ select * from t1 where str='str';
 str
 str
 drop table t1;
+create table t1 (
+str varchar(255) character set utf8 not null,
+key str (str(2))
+) engine=bdb;
+INSERT INTO t1 VALUES ('str');
+INSERT INTO t1 VALUES ('str2');
+select * from t1 where str='str';
+str
+str
+drop table t1;

mysql-test/t/ctype_utf8.test

@@ -294,6 +294,30 @@ select c as c_a from t1 where c='a';
 select c as c_a from t1 where c='б';
 drop table t1;
 
+#
+# Bug 4531: unique key prefix interacts poorly with utf8
+# Check BDB, case insensitive collation
+#
+--disable_warnings
+create table t1 (
+c char(10) character set utf8,
+unique key a (c(1))
+) engine=bdb;
+--enable_warnings
+insert into t1 values ('a'),('b'),('c'),('d'),('e'),('f');
+--error 1062
+insert into t1 values ('aa');
+--error 1062
+insert into t1 values ('aaa');
+insert into t1 values ('б');
+--error 1062
+insert into t1 values ('бб');
+--error 1062
+insert into t1 values ('ббб');
+select c as c_all from t1 order by c;
+select c as c_a from t1 where c='a';
+select c as c_a from t1 where c='б';
+drop table t1;
 
 #
 # Bug 4521: unique key prefix interacts poorly with utf8
@@ -393,6 +417,31 @@ select c as c_a from t1 where c='a';
 select c as c_a from t1 where c='б';
 drop table t1;
 
+#
+# Bug 4531: unique key prefix interacts poorly with utf8
+# Check BDB, binary collation
+#
+--disable_warnings
+create table t1 (
+c char(10) character set utf8 collate utf8_bin,
+unique key a (c(1))
+) engine=bdb;
+--enable_warnings
+insert into t1 values ('a'),('b'),('c'),('d'),('e'),('f');
+--error 1062
+insert into t1 values ('aa');
+--error 1062
+insert into t1 values ('aaa');
+insert into t1 values ('б');
+--error 1062
+insert into t1 values ('бб');
+--error 1062
+insert into t1 values ('ббб');
+select c as c_all from t1 order by c;
+select c as c_a from t1 where c='a';
+select c as c_a from t1 where c='б';
+drop table t1;
+
 
 # Bug#4594: column index make = failed for gbk, but like works
 # Check MYISAM
@@ -429,3 +478,18 @@ INSERT INTO t1 VALUES ('str');
 INSERT INTO t1 VALUES ('str2');
 select * from t1 where str='str';
 drop table t1;
+
+# the same for BDB
+#
+
+--disable_warnings
+create table t1 (
+  str varchar(255) character set utf8 not null,
+  key str (str(2))
+) engine=bdb;
+--enable_warnings
+INSERT INTO t1 VALUES ('str');
+INSERT INTO t1 VALUES ('str2');
+select * from t1 where str='str';
+drop table t1;
+

sql/field.cc

@@ -4382,7 +4382,7 @@ void Field_string::sql_type(String &res) const
 			    (field_length > 3 &&
 			     (table->db_options_in_use &
 			      HA_OPTION_PACK_RECORD) ?
-			      (has_charset() ? "varchar" : "varbinary") : 
+			      (has_charset() ? "varchar" : "varbinary") :
 			      (has_charset() ? "char" : "binary")),
 			    (int) field_length / charset()->mbmaxlen);
   res.length(length);
@@ -4401,6 +4401,24 @@ char *Field_string::pack(char *to, const char *from, uint max_length)
 }
 
 
+char *Field_string::pack_key(char *to, const char *from, uint max_length)
+{
+  const char *end=from+min(field_length,max_length);
+  int length;
+  while (end > from && end[-1] == ' ')
+    end--;
+  length= end-from;
+  uint char_length= (field_charset->mbmaxlen > 1) ?
+              max_length/field_charset->mbmaxlen : max_length;
+  if (length > char_length)
+    char_length= my_charpos(field_charset, from, end, char_length);
+  set_if_smaller(length, char_length);
+  *to= (uchar)length;
+  memcpy(to+1, from, length);
+  return to+1+length;
+}
+
+
 const char *Field_string::unpack(char *to, const char *from)
 {
   uint length= (uint) (uchar) *from++;
@@ -4564,6 +4582,24 @@ char *Field_varstring::pack(char *to, const char *from, uint max_length)
 }
 
 
+char *Field_varstring::pack_key(char *to, const char *from, uint max_length)
+{
+  uint length=uint2korr(from);
+  uint char_length= (field_charset->mbmaxlen > 1) ?
+              max_length/field_charset->mbmaxlen : max_length;
+  from+=HA_KEY_BLOB_LENGTH;
+  if (length > char_length)
+    char_length= my_charpos(field_charset, from, from+length, char_length);
+  set_if_smaller(length, char_length);
+  *to++= (char) (length & 255);
+  if (max_length > 255)
+    *to++= (char) (length >> 8);
+  if (length)
+    memcpy(to, from, length);
+  return to+length;
+}
+
+
 const char *Field_varstring::unpack(char *to, const char *from)
 {
   uint length;
@@ -5139,16 +5175,17 @@ char *Field_blob::pack_key(char *to, const char *from, uint max_length)
   char *save=ptr;
   ptr=(char*) from;
   uint32 length=get_length();			// Length of from string
-  if (length > max_length)
-    length=max_length;
+  uint char_length= (field_charset->mbmaxlen > 1) ?
+              max_length/field_charset->mbmaxlen : max_length;
+  if (length)
+    get_ptr((char**) &from);
+  if (length > char_length)
+    char_length= my_charpos(field_charset, from, from+length, char_length);
+  set_if_smaller(length, char_length);
   *to++= (uchar) length;
   if (max_length > 255)				// 2 byte length
     *to++= (uchar) (length >> 8);
-  if (length)
-  {
-    get_ptr((char**) &from);
-    memcpy(to, from, length);
-  }
+  memcpy(to, from, length);
   ptr=save;					// Restore org row pointer
   return to+length;
 }

sql/field.h

@@ -917,6 +917,7 @@ class Field_string :public Field_str {
   void sort_string(char *buff,uint length);
   void sql_type(String &str) const;
   char *pack(char *to, const char *from, uint max_length=~(uint) 0);
+  char *pack_key(char *to, const char *from, uint max_length);
   const char *unpack(char* to, const char *from);
   int pack_cmp(const char *a,const char *b,uint key_length);
   int pack_cmp(const char *b,uint key_length);
@@ -965,6 +966,7 @@ class Field_varstring :public Field_str {
   void set_key_image(char *buff,uint length, CHARSET_INFO *cs);
   void sql_type(String &str) const;
   char *pack(char *to, const char *from, uint max_length=~(uint) 0);
+  char *pack_key(char *to, const char *from, uint max_length);
   const char *unpack(char* to, const char *from);
   int pack_cmp(const char *a, const char *b, uint key_length);
   int pack_cmp(const char *b, uint key_length);