Bug#30981 CHAR(0x41 USING ucs2) doesn't add leading zero

Bug#30982 CHAR(..USING..) can return a not-well-formed string
Bug#30986 Character set introducer followed by a HEX string can return bad result
check_well_formed_result moved to Item from Item_str_func
fixed Item_func_char::val_str for proper ucs symbols converting
added check for well formed strings for correct conversion of constants with underscore
charset


mysql-test/r/ctype_ucs.result:
  test result
mysql-test/r/ctype_utf8.result:
  test result
mysql-test/t/ctype_ucs.test:
  test case
mysql-test/t/ctype_utf8.test:
  test case
sql/item.cc:
  check_well_formed_result() moved from Item_str_func
sql/item.h:
  check_well_formed_result() moved from Item_str_func
sql/item_strfunc.cc:
  check_well_formed_result moved to Item
  fixed Item_func_char::val_str for proper ucs symbols converting
sql/item_strfunc.h:
  check_well_formed_result moved to Item
sql/sql_yacc.yy:
  added check for well formed string

mysql-test/r/ctype_ucs.result

@@ -922,4 +922,7 @@ ERROR HY000: Illegal mix of collations (ascii_general_ci,IMPLICIT) and (ucs2_gen
 select * from t1 where a=if(b<10,_ucs2 0x0062,_ucs2 0x00C0);
 ERROR HY000: Illegal mix of collations (ascii_general_ci,IMPLICIT) and (ucs2_general_ci,COERCIBLE) for operation '='
 drop table t1;
+select hex(char(0x41 using ucs2));
+hex(char(0x41 using ucs2))
+0041
 End of 5.0 tests

mysql-test/r/ctype_utf8.result

@@ -1538,12 +1538,12 @@ char(53647 using utf8)
 �
 select char(0xff,0x8f using utf8);
 char(0xff,0x8f using utf8)
-ÿ�
+
 Warnings:
 Warning	1300	Invalid utf8 character string: 'FF8F'
 select convert(char(0xff,0x8f) using utf8);
 convert(char(0xff,0x8f) using utf8)
-ÿ�
+
 Warnings:
 Warning	1300	Invalid utf8 character string: 'FF8F'
 set sql_mode=traditional;
@@ -1730,3 +1730,65 @@ i
 1
 н1234567890
 DROP TABLE t1, t2;
+set sql_mode=traditional;
+select hex(char(0xFF using utf8));
+hex(char(0xFF using utf8))
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+select hex(convert(0xFF using utf8));
+hex(convert(0xFF using utf8))
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 0x616263FF);
+hex(_utf8 0x616263FF)
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 X'616263FF');
+hex(_utf8 X'616263FF')
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 B'001111111111');
+hex(_utf8 B'001111111111')
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+select (_utf8 X'616263FF');
+(_utf8 X'616263FF')
+NULL
+Warnings:
+Error	1300	Invalid utf8 character string: 'FF'
+set sql_mode=default;
+select hex(char(0xFF using utf8));
+hex(char(0xFF using utf8))
+
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'
+select hex(convert(0xFF using utf8));
+hex(convert(0xFF using utf8))
+
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 0x616263FF);
+hex(_utf8 0x616263FF)
+616263
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 X'616263FF');
+hex(_utf8 X'616263FF')
+616263
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'
+select hex(_utf8 B'001111111111');
+hex(_utf8 B'001111111111')
+03
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'
+select (_utf8 X'616263FF');
+(_utf8 X'616263FF')
+abc
+Warnings:
+Warning	1300	Invalid utf8 character string: 'FF'

mysql-test/t/ctype_ucs.test

@@ -651,4 +651,9 @@ select * from t1 where a=if(b<10,_ucs2 0x00C0,_ucs2 0x0062);
 select * from t1 where a=if(b<10,_ucs2 0x0062,_ucs2 0x00C0);
 drop table t1;
 
+#
+# Bug#30981 CHAR(0x41 USING ucs2) doesn't add leading zero
+#
+select hex(char(0x41 using ucs2));
+
 --echo End of 5.0 tests

mysql-test/t/ctype_utf8.test

@@ -1403,3 +1403,22 @@ SELECT b FROM t2 UNION SELECT c FROM t1;
 SELECT i FROM t2 UNION SELECT c FROM t1;
 
 DROP TABLE t1, t2;
+
+#
+# Bug#30982: CHAR(..USING..) can return a not-well-formed string
+# Bug #30986: Character set introducer followed by a HEX string can return bad result
+#
+set sql_mode=traditional;
+select hex(char(0xFF using utf8));
+select hex(convert(0xFF using utf8));
+select hex(_utf8 0x616263FF);
+select hex(_utf8 X'616263FF');
+select hex(_utf8 B'001111111111');
+select (_utf8 X'616263FF');
+set sql_mode=default;
+select hex(char(0xFF using utf8));
+select hex(convert(0xFF using utf8));
+select hex(_utf8 0x616263FF);
+select hex(_utf8 X'616263FF');
+select hex(_utf8 B'001111111111');
+select (_utf8 X'616263FF');

sql/item.cc

@@ -4247,6 +4247,41 @@ bool Item::is_datetime()
 }
 
 
+String *Item::check_well_formed_result(String *str)
+{
+  /* Check whether we got a well-formed string */
+  CHARSET_INFO *cs= str->charset();
+  int well_formed_error;
+  uint wlen= cs->cset->well_formed_len(cs,
+                                       str->ptr(), str->ptr() + str->length(),
+                                       str->length(), &well_formed_error);
+  if (wlen < str->length())
+  {
+    THD *thd= current_thd;
+    char hexbuf[7];
+    enum MYSQL_ERROR::enum_warning_level level;
+    uint diff= str->length() - wlen;
+    set_if_smaller(diff, 3);
+    octet2hex(hexbuf, str->ptr() + wlen, diff);
+    if (thd->variables.sql_mode &
+        (MODE_STRICT_TRANS_TABLES | MODE_STRICT_ALL_TABLES))
+    {
+      level= MYSQL_ERROR::WARN_LEVEL_ERROR;
+      null_value= 1;
+      str= 0;
+    }
+    else
+    {
+      level= MYSQL_ERROR::WARN_LEVEL_WARN;
+      str->length(wlen);
+    }
+    push_warning_printf(thd, level, ER_INVALID_CHARACTER_STRING,
+                        ER(ER_INVALID_CHARACTER_STRING), cs->csname, hexbuf);
+  }
+  return str;
+}
+
+
 /*
   Create a field to hold a string value from an item
 

sql/item.h

@@ -870,6 +870,7 @@ public:
   */
   virtual bool result_as_longlong() { return FALSE; }
   bool is_datetime();
+  String *check_well_formed_result(String *str);
 };
 
 

sql/item_strfunc.cc

@@ -38,36 +38,6 @@ C_MODE_END
 String my_empty_string("",default_charset_info);
 
 
-String *Item_str_func::check_well_formed_result(String *str)
-{
-  /* Check whether we got a well-formed string */
-  CHARSET_INFO *cs= str->charset();
-  int well_formed_error;
-  uint wlen= cs->cset->well_formed_len(cs,
-                                       str->ptr(), str->ptr() + str->length(),
-                                       str->length(), &well_formed_error);
-  if (wlen < str->length())
-  {
-    THD *thd= current_thd;
-    char hexbuf[7];
-    enum MYSQL_ERROR::enum_warning_level level;
-    uint diff= str->length() - wlen;
-    set_if_smaller(diff, 3);
-    octet2hex(hexbuf, str->ptr() + wlen, diff);
-    if (thd->variables.sql_mode &
-        (MODE_STRICT_TRANS_TABLES | MODE_STRICT_ALL_TABLES))
-    {
-      level= MYSQL_ERROR::WARN_LEVEL_ERROR;
-      null_value= 1;
-      str= 0;
-    }
-    else
-      level= MYSQL_ERROR::WARN_LEVEL_WARN;
-    push_warning_printf(thd, level, ER_INVALID_CHARACTER_STRING,
-                        ER(ER_INVALID_CHARACTER_STRING), cs->csname, hexbuf);
-  }
-  return str;
-}
 
 
 bool Item_str_func::fix_fields(THD *thd, Item **ref)
@@ -2229,11 +2199,13 @@ String *Item_func_char::val_str(String *str)
 {
   DBUG_ASSERT(fixed == 1);
   str->length(0);
+  str->set_charset(collation.collation);
   for (uint i=0 ; i < arg_count ; i++)
   {
     int32 num=(int32) args[i]->val_int();
     if (!args[i]->null_value)
     {
+      char char_num= (char) num;
       if (num&0xFF000000L) {
         str->append((char)(num>>24));
         goto b2;
@@ -2243,10 +2215,9 @@ String *Item_func_char::val_str(String *str)
       } else if (num&0xFF00L) {
     b1:        str->append((char)(num>>8));
       }
-      str->append((char) num);
+      str->append(&char_num, 1);
     }
   }
-  str->set_charset(collation.collation);
   str->realloc(str->length());			// Add end 0 (for Purify)
   return check_well_formed_result(str);
 }

sql/item_strfunc.h

@@ -35,7 +35,6 @@ public:
   my_decimal *val_decimal(my_decimal *);
   enum Item_result result_type () const { return STRING_RESULT; }
   void left_right_max_length();
-  String *check_well_formed_result(String *str);
   bool fix_fields(THD *thd, Item **ref);
 };
 

sql/sql_yacc.yy

@@ -7716,11 +7716,19 @@ literal:
             String *str= tmp ?
               tmp->quick_fix_field(), tmp->val_str((String*) 0) :
               (String*) 0;
-            $$= new Item_string(str ? str->ptr() : "",
+            $$= new Item_string(NULL, /* name will be set in select_item */
+                                str ? str->ptr() : "",
                                 str ? str->length() : 0,
                                 Lex->underscore_charset);
             if ($$)
+            {
               ((Item_string *) $$)->set_repertoire_from_value();
+              if (!$$->check_well_formed_result(&$$->str_value))
+              {
+                $$= new Item_null();
+                $$->set_name(NULL, 0, system_charset_info);
+              }
+            }
           }
 	| UNDERSCORE_CHARSET BIN_NUM
           {
@@ -7732,9 +7740,15 @@ literal:
 	    String *str= tmp ?
 	      tmp->quick_fix_field(), tmp->val_str((String*) 0) :
 	      (String*) 0;
-	    $$= new Item_string(str ? str->ptr() : "",
-				str ? str->length() : 0,
-				Lex->charset);
+            $$= new Item_string(NULL, /* name will be set in select_item */
+                                str ? str->ptr() : "",
+                                str ? str->length() : 0,
+                                Lex->underscore_charset);
+            if ($$ && !$$->check_well_formed_result(&$$->str_value))
+            {
+              $$= new Item_null();
+              $$->set_name(NULL, 0, system_charset_info);
+            }
           }
 	| DATE_SYM text_literal { $$ = $2; }
 	| TIME_SYM text_literal { $$ = $2; }