Bug #6737: REGEXP gives wrong result with case sensitive collation:

- A new flag MY_CS_CSSORT was introduced for case sensitivity.
- Item_func_regexp doesn't substiture ICASE not only
  for binary collations but for case sensitive collations as well. 

include/m_ctype.h

@@ -63,7 +63,7 @@ typedef struct unicase_info_st
 #define MY_CS_UNICODE	128    /* is a charset is full unicode   */
 #define MY_CS_READY	256    /* if a charset is initialized    */
 #define MY_CS_AVAILABLE	512    /* If either compiled-in or loaded*/
-
+#define MY_CS_CSSORT	1024   /* if case sensitive sort order   */	
 #define MY_CHARSET_UNDEFINED 0
 
 

mysql-test/r/ctype_latin1.result

@@ -296,3 +296,12 @@ FD	C3BD	FD	1
 FE	C3BE	FE	1
 FF	C3BF	FF	1
 DROP TABLE t1;
+select 'a' regexp 'A' collate latin1_general_ci;
+'a' regexp 'A' collate latin1_general_ci
+1
+select 'a' regexp 'A' collate latin1_general_cs;
+'a' regexp 'A' collate latin1_general_cs
+0
+select 'a' regexp 'A' collate latin1_bin;
+'a' regexp 'A' collate latin1_bin
+0

mysql-test/t/ctype_latin1.test

@@ -53,3 +53,10 @@ SELECT
   hex(@l:=convert(@u using latin1)),
   a=@l FROM t1;
 DROP TABLE t1;
+
+#
+# Bug #6737: REGEXP gives wrong result with case sensitive collation
+#
+select 'a' regexp 'A' collate latin1_general_ci;
+select 'a' regexp 'A' collate latin1_general_cs;
+select 'a' regexp 'A' collate latin1_bin;

mysys/charset.c

@@ -228,6 +228,7 @@ static int add_collation(CHARSET_INFO *cs)
       }
       else
       {
+        uchar *sort_order= all_charsets[cs->number]->sort_order;
         simple_cs_init_functions(all_charsets[cs->number]);
         new->mbminlen= 1;
         new->mbmaxlen= 1;
@@ -236,6 +237,16 @@ static int add_collation(CHARSET_INFO *cs)
           all_charsets[cs->number]->state |= MY_CS_LOADED;
         }
         all_charsets[cs->number]->state|= MY_CS_AVAILABLE;
+        
+        /*
+          Check if case sensitive sort order: A < a < B.
+          We need MY_CS_FLAG for regex library, and for
+          case sensitivity flag for 5.0 client protocol,
+          to support isCaseSensitive() method in JDBC driver 
+        */
+        if (sort_order && sort_order['A'] < sort_order['a'] &&
+                          sort_order['a'] < sort_order['B'])
+          all_charsets[cs->number]->state|= MY_CS_CSSORT; 
       }
     }
     else

sql/item_cmpfunc.cc

@@ -2364,8 +2364,9 @@ Item_func_regex::fix_fields(THD *thd, TABLE_LIST *tables, Item **ref)
       return 0;
     }
     int error;
-    if ((error=regcomp(&preg,res->c_ptr(),
-		       (cmp_collation.collation->state & MY_CS_BINSORT) ?
+    if ((error= regcomp(&preg,res->c_ptr(),
+                        ((cmp_collation.collation->state & MY_CS_BINSORT) ||
+                         (cmp_collation.collation->state & MY_CS_CSSORT)) ?
                          REG_EXTENDED | REG_NOSUB :
                          REG_EXTENDED | REG_NOSUB | REG_ICASE,
                         cmp_collation.collation)))
@@ -2416,7 +2417,8 @@ longlong Item_func_regex::val_int()
 	regex_compiled=0;
       }
       if (regcomp(&preg,res2->c_ptr(),
-		  (cmp_collation.collation->state & MY_CS_BINSORT) ?
+                  ((cmp_collation.collation->state & MY_CS_BINSORT) ||
+                   (cmp_collation.collation->state & MY_CS_CSSORT)) ?
                    REG_EXTENDED | REG_NOSUB :
                    REG_EXTENDED | REG_NOSUB | REG_ICASE,
                    cmp_collation.collation))

strings/ctype-czech.c

@@ -590,7 +590,7 @@ static MY_COLLATION_HANDLER my_collation_latin2_czech_ci_handler =
 CHARSET_INFO my_charset_latin2_czech_ci =
 {
     2,0,0,                                      /* number    */
-    MY_CS_COMPILED|MY_CS_STRNXFRM,	/* state      */
+    MY_CS_COMPILED|MY_CS_STRNXFRM|MY_CS_CSSORT, /* state     */
     "latin2",                                   /* cs name   */
     "latin2_czech_cs",                          /* name      */
     "",                                         /* comment   */

strings/ctype-win1250ch.c

@@ -625,7 +625,7 @@ static MY_COLLATION_HANDLER my_collation_czech_ci_handler =
 CHARSET_INFO my_charset_cp1250_czech_ci =
 {
   34,0,0,                                     /* number    */
-  MY_CS_COMPILED|MY_CS_STRNXFRM,		/* state     */
+  MY_CS_COMPILED|MY_CS_STRNXFRM|MY_CS_CSSORT, /* state     */
   "cp1250",                                   /* cs name   */
   "cp1250_czech_cs",                          /* name      */
   "",                                         /* comment   */