Don't allow badly formed UTF8 identitiers

include/mysqld_error.h

@@ -316,4 +316,5 @@
 #define ER_GET_TEMPORARY_ERRMSG 1297
 #define ER_UNKNOWN_TIME_ZONE 1298
 #define ER_WARN_INVALID_TIMESTAMP 1299
-#define ER_ERROR_MESSAGES 300
+#define ER_INVALID_CHARACTER_STRING 1300
+#define ER_ERROR_MESSAGES 301

mysql-test/r/ctype_recoding.result

@@ -136,3 +136,9 @@ SET character_set_connection=binary;
 SELECT 'тест' as s;
 s
 тест
+SET NAMES binary;
+CREATE TABLE `good` (a int);
+ERROR HY000: Invalid utf8 character string: ''
+SET NAMES utf8;
+CREATE TABLE `good` (a int);
+ERROR HY000: Invalid utf8 character string: '` (a int)'

mysql-test/t/ctype_recoding.test

@@ -97,3 +97,13 @@ SELECT 'тест' as s;
 SET NAMES utf8;
 SET character_set_connection=binary;
 SELECT 'тест' as s;
+
+#
+# Test that we allow only well-formed UTF8 identitiers
+#
+SET NAMES binary;
+--error 1300
+CREATE TABLE `good` (a int);
+SET NAMES utf8;
+--error 1300
+CREATE TABLE `good` (a int);

sql/share/czech/errmsg.txt

@@ -312,3 +312,4 @@ character-set=latin2
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/danish/errmsg.txt

@@ -306,3 +306,4 @@ character-set=latin1
 "Modtog temporary fejl %d '%-.100s' fra %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/dutch/errmsg.txt

@@ -314,3 +314,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/english/errmsg.txt

@@ -303,3 +303,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/estonian/errmsg.txt

@@ -308,3 +308,4 @@ character-set=latin7
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/french/errmsg.txt

@@ -303,3 +303,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/german/errmsg.txt

@@ -315,3 +315,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/greek/errmsg.txt

@@ -303,3 +303,4 @@ character-set=greek
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/hungarian/errmsg.txt

@@ -305,3 +305,4 @@ character-set=latin2
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/italian/errmsg.txt

@@ -303,3 +303,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/japanese/errmsg.txt

@@ -305,3 +305,4 @@ character-set=ujis
 "Got temporary NDB error %d '%-.100s'",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/korean/errmsg.txt

@@ -303,3 +303,4 @@ character-set=euckr
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/norwegian-ny/errmsg.txt

@@ -305,3 +305,4 @@ character-set=latin1
 "Mottok temporary feil %d '%-.100s' fra %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/norwegian/errmsg.txt

@@ -305,3 +305,4 @@ character-set=latin1
 "Mottok temporary feil %d '%-.100s' fra %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/polish/errmsg.txt

@@ -307,3 +307,4 @@ character-set=latin2
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/portuguese/errmsg.txt

@@ -304,3 +304,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/romanian/errmsg.txt

@@ -307,3 +307,4 @@ character-set=latin2
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/russian/errmsg.txt

@@ -305,3 +305,4 @@ character-set=koi8r
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/serbian/errmsg.txt

@@ -309,3 +309,4 @@ character-set=cp1250
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/slovak/errmsg.txt

@@ -311,3 +311,4 @@ character-set=latin2
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/spanish/errmsg.txt

@@ -305,3 +305,4 @@ character-set=latin1
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/swedish/errmsg.txt

@@ -303,3 +303,4 @@ character-set=latin1
 "Fick tilfällig felkod %d '%-.100s' från %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/share/ukrainian/errmsg.txt

@@ -308,3 +308,4 @@ character-set=koi8u
 "Got temporary error %d '%-.100s' from %s",
 "Unknown or incorrect time zone: '%-.64s'",
 "Invalid TIMESTAMP value in column '%s' at row %ld",
+"Invalid %s character string: '%.64s'",

sql/sql_yacc.yy

@@ -4884,7 +4884,19 @@ IDENT_sys:
 	  {
 	    THD *thd= YYTHD;
 	    if (thd->charset_is_system_charset)
+            {
+              CHARSET_INFO *cs= system_charset_info;
+              uint wlen= cs->cset->well_formed_len(cs, $1.str,
+                                                   $1.str+$1.length,
+                                                   $1.length);
+              if (wlen < $1.length)
+              {
+                net_printf(YYTHD, ER_INVALID_CHARACTER_STRING, cs->csname,
+                           $1.str + wlen);
+                YYABORT;
+              }
 	      $$= $1;
+            }
 	    else
 	      thd->convert_string(&$$, system_charset_info,
 				  $1.str, $1.length, thd->charset());