Bug#24660: "enum" field type definition problem

ENUMs weren't allowed to have character 0xff, a perfectly good
character in some locales.  This was circumvented by mapping 0xff in
ENUMs to ',', thereby prevent actual commas from being used. Now if
0xff makes an appearance, we find a character not used in the enum and
use that as a separator. If no such character exists, we throw an
error.

Any solution would have broken some sort of existing behaviour. This
solution should serve both fractions (those with 0xff and those with
',' in their enums), but WILL REQUIRE A DUMP/RESTORE CYCLE FROM THOSE
WITH 0xff IN THEIR ENUMS. :-/ That is, mysqldump with their current
server, and restore when upgrading to one with this patch.

mysql-test/r/type_enum.result

@@ -1754,3 +1754,29 @@ t1	CREATE TABLE `t1` (
   `f2` enum('') DEFAULT NULL
 ) ENGINE=MyISAM DEFAULT CHARSET=latin1
 drop table t1;
+End of 4.1 tests
+create table t1(russian enum('E','F','EF','FE') NOT NULL DEFAULT'E');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `russian` enum('E','F','EF','FE') NOT NULL DEFAULT 'E'
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+drop table t1;
+create table t1(denormal enum('E','F','E,F','F,E') NOT NULL DEFAULT'E');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `denormal` enum('E','F','E,F','F,E') NOT NULL DEFAULT 'E'
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+drop table t1;
+create table t1(russian_deviant enum('E','F','EF','F,E') NOT NULL DEFAULT'E');
+show create table t1;
+Table	Create Table
+t1	CREATE TABLE `t1` (
+  `russian_deviant` enum('E','F','EF','F,E') NOT NULL DEFAULT 'E'
+) ENGINE=MyISAM DEFAULT CHARSET=latin1
+drop table t1;
+create table t1(exhausting_charset enum('ABCDEFGHIJKLMNOPQRSTUVWXYZ','
	
+
 !"','#$%&\'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~','xx\','yy\','zz'));
+ERROR 42000: Field separator argument is not what is expected; check the manual
+End of 5.1 tests

mysql-test/t/type_enum.test

@@ -136,4 +136,26 @@ alter table t1 add f2 enum(0xFFFF);
 show create table t1;
 drop table t1;
 
-# End of 4.1 tests
+--echo End of 4.1 tests
+
+#
+# Bug#24660 "enum" field type definition problem
+#
+create table t1(russian enum('E','F','EF','FE') NOT NULL DEFAULT'E');
+show create table t1;
+drop table t1;
+
+create table t1(denormal enum('E','F','E,F','F,E') NOT NULL DEFAULT'E');
+show create table t1;
+drop table t1;
+
+create table t1(russian_deviant enum('E','F','EF','F,E') NOT NULL DEFAULT'E');
+show create table t1;
+drop table t1;
+
+# ER_WRONG_FIELD_TERMINATORS
+--error 1083
+create table t1(exhausting_charset enum('ABCDEFGHIJKLMNOPQRSTUVWXYZ','
	
+
 !"','#$%&\'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\\]^_`abcdefghijklmnopqrstuvwxyz{|}~','xx\','yy\','zz'));
+
+--echo End of 5.1 tests

sql/table.cc

@@ -803,17 +803,6 @@ static int open_binary_frm(THD *thd, TABLE_SHARE *share, uchar *head,
       {
         char *val= (char*) interval->type_names[count];
         interval->type_lengths[count]= strlen(val);
-        /*
-          Replace all ',' symbols with NAMES_SEP_CHAR.
-          See the comment in unireg.cc, pack_fields() function
-          for details.
-        */
-        for (uint cnt= 0 ; cnt < interval->type_lengths[count] ; cnt++)
-        {
-          char c= val[cnt];
-          if (c == ',')
-            val[cnt]= NAMES_SEP_CHAR;
-        }       
       }
       interval->type_lengths[count]= 0;
     }

sql/unireg.cc

@@ -788,27 +788,46 @@ static bool pack_fields(File file, List<create_field> &create_fields,
     {
       if (field->interval_id > int_count)
       {
-	int_count=field->interval_id;
-	tmp.append(NAMES_SEP_CHAR);
-	for (const char **pos=field->interval->type_names ; *pos ; pos++)
+        unsigned char  sep= 0;
+        unsigned char  occ[256];
+        uint           i;
+        unsigned char *val= NULL;
+
+        bzero(occ, sizeof(occ));
+
+        for (i=0; (val= (unsigned char*) field->interval->type_names[i]); i++)
+          for (uint j = 0; j < field->interval->type_lengths[i]; j++)
+            occ[(unsigned int) (val[j])]= 1;
+
+        if (!occ[(unsigned char)NAMES_SEP_CHAR])
+          sep= (unsigned char) NAMES_SEP_CHAR;
+        else if (!occ[(unsigned int)','])
+          sep= ',';
+        else
         {
-          char *val= (char*) *pos;
-          uint str_len= strlen(val);
-          /*
-            Note, hack: in old frm NAMES_SEP_CHAR is used to separate
-            names in the interval (ENUM/SET). To allow names to contain
-            NAMES_SEP_CHAR, we replace it with a comma before writing frm.
-            Backward conversion is done during frm file opening,
-            See table.cc, openfrm() function
-          */
-          for (uint cnt= 0 ; cnt < str_len ; cnt++)
+          for (uint i=1; i<256; i++)
           {
-            char c= val[cnt];
-            if (c == NAMES_SEP_CHAR)
-              val[cnt]= ',';
+            if(!occ[i])
+            {
+              sep= i;
+              break;
+            }
           }
+
+          if(!sep)    /* disaster, enum uses all characters, none left as separator */
+          {
+            my_message(ER_WRONG_FIELD_TERMINATORS,ER(ER_WRONG_FIELD_TERMINATORS),
+                       MYF(0));
+            DBUG_RETURN(1);
+          }
+        }
+
+        int_count= field->interval_id;
+        tmp.append(sep);
+        for (const char **pos=field->interval->type_names ; *pos ; pos++)
+        {
           tmp.append(*pos);
-	  tmp.append(NAMES_SEP_CHAR);
+          tmp.append(sep);
         }
         tmp.append('\0');                      // End of intervall
       }