Simple charsets now have strncoll() field too

client/sql_string.cc

@@ -507,7 +507,7 @@ int sortcmp(const String *x,const String *y)
   uint32 x_len=x->length(),y_len=y->length(),len=min(x_len,y_len);
 
 #ifdef USE_STRCOLL
-  if (use_strcoll(x->str_charset))
+  if (use_strnxfrm(x->str_charset))
   {
 #ifndef CMP_ENDSPACE
     while (x_len && my_isspace(x->str_charset,s[x_len-1]))

include/m_ctype.h

@@ -92,11 +92,12 @@ typedef struct charset_info_st
   int (*wc_mb)(struct charset_info_st *cs,my_wc_t wc,
 	       unsigned char *s,unsigned char *e);
     
-  /* Functions for case convertion */
+  /* Functions for case and sort convertion */
   void    (*caseup_str)(struct charset_info_st *, char *);
   void    (*casedn_str)(struct charset_info_st *, char *);
   void    (*caseup)(struct charset_info_st *, char *, uint);
   void    (*casedn)(struct charset_info_st *, char *, uint);
+  void    (*tosort)(struct charset_info_st *, char *, uint);
   
   /* Functions for case comparison */
   int  (*strcasecmp)(struct charset_info_st *, const char *, const char *);
@@ -125,14 +126,15 @@ extern const char *compiled_charset_name(uint charset_number);
 #define MY_CHARSET_CURRENT (default_charset_info->number)
 
 /* declarations for simple charsets */
-extern int  my_strnxfrm_simple(CHARSET_INFO *, char *, uint, const char *, uint); 
-extern int  my_strnncoll_simple(CHARSET_INFO *, const char *, uint, const char *, uint);
+extern int  my_strnxfrm_simple(CHARSET_INFO *, uchar *, uint, const uchar *, uint); 
+extern int  my_strnncoll_simple(CHARSET_INFO *, const uchar *, uint, const uchar *, uint);
 
 /* Functions for 8bit */
 extern void my_caseup_str_8bit(CHARSET_INFO *, char *);
 extern void my_casedn_str_8bit(CHARSET_INFO *, char *);
 extern void my_caseup_8bit(CHARSET_INFO *, char *, uint);
 extern void my_casedn_8bit(CHARSET_INFO *, char *, uint);
+extern void my_tosort_8bit(CHARSET_INFO *, char *, uint);
 
 extern int my_strcasecmp_8bit(CHARSET_INFO * cs, const char *, const char *);
 extern int my_strncasecmp_8bit(CHARSET_INFO * cs, const char *, const char *, uint);
@@ -364,7 +366,7 @@ void my_hash_sort_ucs2(struct charset_info_st *cs, const uchar *key, uint len, u
 #define my_isvar(s,c)                 (my_isalnum(s,c) || (c) == '_')
 #define my_isvar_start(s,c)           (my_isalpha(s,c) || (c) == '_')
 
-#define use_strcoll(s)                ((s)->strnncoll != NULL)
+#define use_strnxfrm(s)               ((s)->strnxfrm  != NULL)
 #define my_strnxfrm(s, a, b, c, d)    ((s)->strnxfrm((s), (a), (b), (c), (d)))
 #define my_strnncoll(s, a, b, c, d)   ((s)->strnncoll((s), (a), (b), (c), (d)))
 #define my_like_range(s, a, b, c, d, e, f, g, h) \
@@ -377,6 +379,7 @@ void my_hash_sort_ucs2(struct charset_info_st *cs, const uchar *key, uint len, u
 
 #define my_caseup(s, a, l)            ((s)->caseup((s), (a), (l)))
 #define my_casedn(s, a, l)            ((s)->casedn((s), (a), (l)))
+#define my_tosort(s, a, l)            ((s)->tosort((s), (a), (l)))
 #define my_caseup_str(s, a)           ((s)->caseup_str((s), (a)))
 #define my_casedn_str(s, a)           ((s)->casedn_str((s), (a)))
 #define my_strcasecmp(s, a, b)        ((s)->strcasecmp((s), (a), (b)))

include/my_sys.h

@@ -583,7 +583,6 @@ extern void allow_break(void);
 #endif
 
 extern void my_remember_signal(int signal_number,sig_handler (*func)(int));
-extern void case_sort(CHARSET_INFO *cs, my_string str,uint length);
 extern uint dirname_part(my_string to,const char *name);
 extern uint dirname_length(const char *name);
 #define base_name(A) (A+dirname_length(A))

isam/_key.c

@@ -54,11 +54,11 @@ uint _nisam_make_key(register N_INFO *info, uint keynr, uchar *key, const char *
       *key++= (uchar) (length=(uint) (end-pos));
       memcpy((byte*) key,(byte*) pos,(size_t) length);
 #ifdef USE_STRCOLL
-      if (!use_strcoll(default_charset_info))
+      if (!use_strnxfrm(default_charset_info))
 #endif
       {
 	if (type == HA_KEYTYPE_TEXT)
-	  case_sort(default_charset_info,(byte*) key,length);
+	  my_tosort(default_charset_info,(byte*) key,length);
       }
       key+=length;
     }
@@ -67,11 +67,11 @@ uint _nisam_make_key(register N_INFO *info, uint keynr, uchar *key, const char *
       memcpy((byte*) key,(byte*) record+keyseg->base.start,
 	     (size_t) keyseg->base.length);
 #ifdef USE_STRCOLL
-      if (!use_strcoll(default_charset_info))
+      if (!use_strnxfrm(default_charset_info))
 #endif
       {
 	if (type == HA_KEYTYPE_TEXT)
-	  case_sort(default_charset_info,(byte*) key,(uint) keyseg->base.length);
+	  my_tosort(default_charset_info,(byte*) key,(uint) keyseg->base.length);
       }
 #ifdef NAN_TEST
       else if (type == HA_KEYTYPE_FLOAT)
@@ -150,11 +150,11 @@ uint _nisam_pack_key(register N_INFO *info, uint keynr, uchar *key, uchar *old,
     else
       memcpy((byte*) key,old,(size_t) length);
 #ifdef USE_STRCOLL
-      if (!use_strcoll(default_charset_info))
+      if (!use_strnxfrm(default_charset_info))
 #endif
       {
 	if (type == HA_KEYTYPE_TEXT)
-	  case_sort(default_charset_info,(byte*) key,length);
+	  my_tosort(default_charset_info,(byte*) key,length);
       }
       key+= length;
   }

isam/_search.c

@@ -332,7 +332,7 @@ int _nisam_key_cmp(register N_KEYSEG *keyseg, register uchar *a, register uchar
 	end= a+ min(key_length,(uint) length);
 
 #ifdef USE_STRCOLL
-        if (use_strcoll(default_charset_info)) {
+        if (use_strnxfrm(default_charset_info)) {
           if (((enum ha_base_keytype) keyseg->base.type) == HA_KEYTYPE_BINARY)
           {
             while (a < end)
@@ -383,7 +383,7 @@ int _nisam_key_cmp(register N_KEYSEG *keyseg, register uchar *a, register uchar
       else
       {
 #ifdef USE_STRCOLL
-        if (use_strcoll(default_charset_info)) {
+        if (use_strnxfrm(default_charset_info)) {
           if (((enum ha_base_keytype) keyseg->base.type) == HA_KEYTYPE_BINARY)
           {
             while (a < end)

myisam/mi_open.c

@@ -685,7 +685,7 @@ static void setup_key_functions(register MI_KEYDEF *keyinfo)
     keyinfo->get_key= _mi_get_pack_key;
     if (keyinfo->seg[0].flag & HA_PACK_KEY)
     {						/* Prefix compression */
-      if (!keyinfo->seg->charset || use_strcoll(keyinfo->seg->charset) ||
+      if (!keyinfo->seg->charset || use_strnxfrm(keyinfo->seg->charset) ||
           (keyinfo->seg->flag & HA_NULL_PART))
         keyinfo->bin_search=_mi_seq_search;
       else

myisam/mi_search.c

@@ -1287,7 +1287,7 @@ _mi_calc_var_pack_key_length(MI_KEYDEF *keyinfo,uint nod_flag,uchar *next_key,
   if ((keyinfo->flag & HA_FULLTEXT) &&
       ((keyseg->type == HA_KEYTYPE_TEXT) ||
        (keyseg->type == HA_KEYTYPE_VARTEXT)) &&
-      !use_strcoll(keyseg->charset))
+      !use_strnxfrm(keyseg->charset))
     sort_order=keyseg->charset->sort_order;
 
   /* diff flag contains how many bytes is needed to pack key */

mysys/charset.c

@@ -377,6 +377,7 @@ static CHARSET_INFO *add_charset(uint cs_number, myf flags)
   cs->casedn_str  = my_casedn_str_8bit;
   cs->caseup      = my_caseup_8bit;
   cs->casedn      = my_casedn_8bit;
+  cs->tosort      = my_tosort_8bit;
   cs->strcasecmp  = my_strcasecmp_8bit;
   cs->strncasecmp = my_strncasecmp_8bit;
   cs->mb_wc       = my_mb_wc_8bit;

mysys/mf_casecnv.c

@@ -15,11 +15,7 @@
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA */
 
 /*
-  Functions to convert to lover_case and to upper_case in scandinavia.
-
-  case_sort converts a character string to a representaion that can
-  be compared by strcmp to find with is alfabetical bigger.
-  (lower- and uppercase letters is compared as the same)
+  Functions to convert to lover_case and to upper_case.
 */
 
 #include "mysys_priv.h"
@@ -29,21 +25,11 @@
 #endif
 
 
-	/* to sort-string that can be compared to get text in order */
-
-void case_sort(CHARSET_INFO *cs, my_string str, uint length)
-{
-  register uchar *map=cs->sort_order;
-  
-  for ( ; length>0 ; length--, str++)
-    *str= (char) map[(uchar) *str];
-} /* case_sort */
-
 
 int my_sortcmp(CHARSET_INFO *cs, const char *s, const char *t, uint len)
 {
 #ifdef USE_STRCOLL
-  if (use_strcoll(cs))
+  if (use_strnxfrm(cs))
     return my_strnncoll(cs,(uchar *)s, len, (uchar *)t, len);
   else
 #endif
@@ -63,7 +49,7 @@ int my_sortncmp(CHARSET_INFO *cs,
                 const char *t, uint t_len)
 {
 #ifdef USE_STRCOLL
-  if (use_strcoll(cs))
+  if (use_strnxfrm(cs))
     return my_strnncoll(cs, (uchar *)s, s_len, (uchar *)t, t_len);
   else
 #endif

mysys/my_handler.c

@@ -23,7 +23,7 @@ int mi_compare_text(CHARSET_INFO *charset_info, uchar *a, uint a_length,
   int flag;
 
 #ifdef USE_STRCOLL
-  if (use_strcoll(charset_info))
+  if (use_strnxfrm(charset_info))
   {
     if (part_key && b_length < a_length)
       a_length=b_length;

sql/field.cc

@@ -3841,7 +3841,7 @@ void Field_string::sort_string(char *to,uint length)
   else
   {
 #ifdef USE_STRCOLL
-    if (use_strcoll(field_charset)) {
+    if (use_strnxfrm(field_charset)) {
       uint tmp=my_strnxfrm(field_charset,
                           (unsigned char *)to, length,
                           (unsigned char *) ptr, field_length);
@@ -4045,7 +4045,7 @@ void Field_varstring::sort_string(char *to,uint length)
   else
   {
 #ifdef USE_STRCOLL
-    if (use_strcoll(field_charset))
+    if (use_strnxfrm(field_charset))
       tot_length=my_strnxfrm(field_charset,
                              (unsigned char *) to, length,
                              (unsigned char *)ptr+2, tot_length);
@@ -4536,7 +4536,7 @@ void Field_blob::sort_string(char *to,uint length)
     else
     {
 #ifdef USE_STRCOLL
-      if (use_strcoll(field_charset))
+      if (use_strnxfrm(field_charset))
       {
         blob_length=my_strnxfrm(field_charset,
                                 (unsigned char *)to, length, 

sql/filesort.cc

@@ -140,7 +140,7 @@ ha_rows filesort(TABLE *table, SORT_FIELD *sortorder, uint s_length,
     records=param.max_rows;			/* purecov: inspected */
 
 #ifdef USE_STRCOLL
-  if (use_strcoll(charset) &&
+  if (use_strnxfrm(charset) &&
       !(param.tmp_buffer=my_malloc(param.sort_length,MYF(MY_WME))))
     goto err;
 #endif
@@ -511,7 +511,7 @@ static void make_sortkey(register SORTPARAM *param,
 	    length=sort_field->length;
 	  }
 #ifdef USE_STRCOLL
-          if(use_strcoll(cs))
+          if(use_strnxfrm(cs))
           {
             if (item->binary)
             {
@@ -541,7 +541,7 @@ static void make_sortkey(register SORTPARAM *param,
               memcpy(to,res->ptr(),length);
             bzero((char *)to+length,diff);
             if (!item->binary)
-              case_sort(cs, (char*) to,length);
+              my_tosort(cs, (char*) to,length);
 #ifdef USE_STRCOLL
           }
 #endif
@@ -946,7 +946,7 @@ sortlength(SORT_FIELD *sortorder, uint s_length)
 	if (!sortorder->field->binary())
 	{
 	  CHARSET_INFO *cs=((Field_str*)(sortorder->field))->charset();
-	  if (use_strcoll(cs))
+	  if (use_strnxfrm(cs))
 	    sortorder->length= sortorder->length*cs->strxfrm_multiply;
 	}
 #endif
@@ -966,7 +966,7 @@ sortlength(SORT_FIELD *sortorder, uint s_length)
 	if (!sortorder->item->binary)
 	{ 
 	  CHARSET_INFO *cs=sortorder->item->str_value.charset();
-	  if (use_strcoll(cs))
+	  if (use_strnxfrm(cs))
 	    sortorder->length= sortorder->length*cs->strxfrm_multiply;
 	}
 #endif

sql/opt_range.cc

@@ -978,7 +978,7 @@ get_mm_leaf(PARAM *param, Field *field, KEY_PART *key_part,
     {
       CHARSET_INFO *charset=((Field_str*)(field))->charset();
 #ifdef USE_STRCOLL
-      if (use_strcoll(charset))
+      if (use_strnxfrm(charset))
         like_error= my_like_range(charset,
                                   res->ptr(),res->length(),wild_prefix,
                                   field_length, min_str+maybe_null,

sql/sql_string.cc

@@ -505,7 +505,7 @@ int sortcmp(const String *x,const String *y)
   uint32 x_len=x->length(),y_len=y->length(),len=min(x_len,y_len);
 
 #ifdef USE_STRCOLL
-  if (use_strcoll(x->str_charset))
+  if (use_strnxfrm(x->str_charset))
   {
 #ifndef CMP_ENDSPACE
     while (x_len && my_isspace(x->str_charset,s[x_len-1]))

strings/ctype-simple.c

@@ -20,32 +20,30 @@
 #include "assert.h"
 
 int my_strnxfrm_simple(CHARSET_INFO * cs, 
-                       char *dest, uint len,
-                       const char *src, uint srclen)
+                       uchar *dest, uint len,
+                       const uchar *src, uint srclen)
 {
   uchar *map= cs->sort_order;
   DBUG_ASSERT(len >= srclen);
   
   for ( ; len > 0 ; len-- )
-    *dest++= (char) map[(uchar) *src++];
+    *dest++= map[*src++];
   return srclen;
 }
 
-int my_strnncoll_simple(CHARSET_INFO * cs, const char *s, uint slen, 
-			const char *t, uint tlen)
+int my_strnncoll_simple(CHARSET_INFO * cs, const uchar *s, uint slen, 
+			const uchar *t, uint tlen)
 {
   int len = ( slen > tlen ) ? tlen : slen;
   uchar *map= cs->sort_order;
   while (len--)
   {
-    if (map[(uchar) *s++] != map[(uchar) *t++])
-      return ((int) map[(uchar) s[-1]] -
-              (int) map[(uchar) t[-1]]);
+    if (map[*s++] != map[*t++])
+      return ((int) map[s[-1]] - (int) map[t[-1]]);
   }
   return (int) (slen-tlen);
 }
 
-
 void my_caseup_str_8bit(CHARSET_INFO * cs,char *str)
 {
   register uchar *map=cs->to_upper;
@@ -53,7 +51,6 @@ void my_caseup_str_8bit(CHARSET_INFO * cs,char *str)
     str++;
 }
 
-
 void my_casedn_str_8bit(CHARSET_INFO * cs,char *str)
 {
   register uchar *map=cs->to_lower;
@@ -61,7 +58,6 @@ void my_casedn_str_8bit(CHARSET_INFO * cs,char *str)
     str++;
 }
 
-
 void my_caseup_8bit(CHARSET_INFO * cs, char *str, uint length)
 {
   register uchar *map=cs->to_upper;
@@ -76,6 +72,12 @@ void my_casedn_8bit(CHARSET_INFO * cs, char *str, uint length)
     *str= (char) map[(uchar) *str];
 }
 
+void my_tosort_8bit(CHARSET_INFO *cs, char *str, uint length)
+{
+  register uchar *map=cs->sort_order;
+  for ( ; length>0 ; length--, str++)
+    *str= (char) map[(uchar) *str];
+}
 
 int my_strcasecmp_8bit(CHARSET_INFO * cs,const char *s, const char *t)
 {