Fix for problem that GROUP BY and DISTINCT didn't work for UTF8

heap/hp_hash.c

@@ -151,7 +151,8 @@ void _hp_movelink(HASH_INFO *pos, HASH_INFO *next_link, HASH_INFO *newlink)
 
 ulong _hp_hashnr(register HP_KEYDEF *keydef, register const byte *key)
 {
-  register ulong nr=1, nr2=4;
+  /*register*/ 
+  ulong nr=1, nr2=4;
   HP_KEYSEG *seg,*endseg;
 
   for (seg=keydef->seg,endseg=seg+keydef->keysegs ; seg < endseg ; seg++)
@@ -170,6 +171,10 @@ ulong _hp_hashnr(register HP_KEYDEF *keydef, register const byte *key)
     }
     if (seg->type == HA_KEYTYPE_TEXT)
     {
+      if (default_charset_info->hash_sort)
+        default_charset_info->hash_sort(default_charset_info,
+        				pos,((uchar*)key)-pos,&nr,&nr2);
+      else
       for (; pos < (uchar*) key ; pos++)
       {
 	nr^=(ulong) ((((uint) nr & 63)+nr2) * 
@@ -193,7 +198,8 @@ ulong _hp_hashnr(register HP_KEYDEF *keydef, register const byte *key)
 
 ulong _hp_rec_hashnr(register HP_KEYDEF *keydef, register const byte *rec)
 {
-  register ulong nr=1, nr2=4;
+  /*register*/
+  ulong nr=1, nr2=4;
   HP_KEYSEG *seg,*endseg;
 
   for (seg=keydef->seg,endseg=seg+keydef->keysegs ; seg < endseg ; seg++)
@@ -209,6 +215,10 @@ ulong _hp_rec_hashnr(register HP_KEYDEF *keydef, register const byte *rec)
     }
     if (seg->type == HA_KEYTYPE_TEXT)
     {
+      if (default_charset_info->hash_sort)
+        default_charset_info->hash_sort(default_charset_info,
+        				pos,end-pos,&nr,&nr2);
+      else
       for (; pos < end ; pos++)
       {
 	nr^=(ulong) ((((uint) nr & 63)+nr2)*

include/m_ctype.h

@@ -94,6 +94,8 @@ typedef struct charset_info_st
     
     /* Hash calculation */
     uint (*hash_caseup)(struct charset_info_st *cs, const byte *key, uint len);
+    void (*hash_sort)(struct charset_info_st *cs, const uchar *key, uint len, ulong *nr1, ulong *nr2); 
+    
     char    max_sort_char; /* For LIKE otimization */
 } CHARSET_INFO;
 
@@ -279,7 +281,8 @@ int my_utf8_uni (CHARSET_INFO *cs, my_wc_t *p, const uchar *s, const uchar *e);
 int my_uni_utf8 (CHARSET_INFO *cs, my_wc_t wc, uchar *b, uchar *e);
 
 uint my_hash_caseup_utf8(struct charset_info_st *cs, const byte *key, uint len);
-
+void my_hash_sort_utf8(struct charset_info_st *cs, const uchar *key, uint len, ulong *nr1, ulong *nr2); 
+   
 #endif
 
 #define	_U	01	/* Upper case */

strings/ctype-utf8.c

@@ -1754,6 +1754,26 @@ uint my_hash_caseup_utf8(CHARSET_INFO *cs, const byte *s, uint slen)
   return nr;
 }
 
+
+void my_hash_sort_utf8(CHARSET_INFO *cs, const uchar *s, uint slen, ulong *n1, ulong *n2)
+{
+  my_wc_t wc;
+  int res;
+  const uchar *e=s+slen;
+
+  while ((s < e) && (res=my_utf8_uni(cs,&wc, (uchar *)s, (uchar*)e))>0 )
+  {
+    int plane = (wc>>8) & 0xFF;
+    wc = uni_plane[plane] ? uni_plane[plane][wc & 0xFF].sort : wc;
+    n1[0]^= (((n1[0] & 63)+n2[0])*(wc & 0xFF))+ (n1[0] << 8);
+    n2[0]+=3;
+    n1[0]^= (((n1[0] & 63)+n2[0])*(wc >> 8))+ (n1[0] << 8);
+    n2[0]+=3;
+    s+=res;
+  }
+}
+
+
 void my_caseup_str_utf8(CHARSET_INFO * cs, char * s)
 {
   my_caseup_utf8(cs, s, strlen(s));

strings/ctype.c

@@ -2830,6 +2830,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -2861,6 +2862,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_mb,
     my_strncasecmp_mb,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -2892,6 +2894,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -2923,6 +2926,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -2954,6 +2958,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -2985,6 +2990,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3016,6 +3022,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3047,6 +3054,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3078,6 +3086,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3109,6 +3118,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3140,6 +3150,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_mb,
     my_strncasecmp_mb,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3171,6 +3182,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_mb,
     my_strncasecmp_mb,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3202,6 +3214,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_mb,
     my_strncasecmp_mb,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3233,6 +3246,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3264,6 +3278,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3295,6 +3310,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3326,6 +3342,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3357,6 +3374,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3388,6 +3406,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3419,6 +3438,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3450,6 +3470,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3481,6 +3502,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3512,6 +3534,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3543,6 +3566,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3574,6 +3598,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3605,6 +3630,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3636,6 +3662,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_mb,
     my_strncasecmp_mb,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3667,6 +3694,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_utf8,
     my_strncasecmp_utf8,
     my_hash_caseup_utf8,/* hash_caseup */
+    my_hash_sort_utf8,	/* hash_sort   */
     0
   },
 #endif
@@ -3698,6 +3726,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3729,6 +3758,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3760,6 +3790,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3791,6 +3822,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3822,6 +3854,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3853,6 +3886,7 @@ CHARSET_INFO compiled_charsets[] = {
     my_strcasecmp_8bit,
     my_strncasecmp_8bit,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   },
 #endif
@@ -3885,6 +3919,7 @@ CHARSET_INFO compiled_charsets[] = {
     NULL,
     NULL,
     NULL,		/* hash_caseup */
+    NULL,		/* hash_sort   */
     0
   }
 };