CHARSET_INFO::instr was extended to return more substring match results:

- offset of substr begining
- offset of substr end
- number of characters (MB compatible)

include/m_ctype.h

@@ -75,6 +75,12 @@ typedef struct my_uni_idx_st
   uchar  *tab;
 } MY_UNI_IDX;
 
+typedef struct
+{
+  uint beg;
+  uint end;
+  uint mblen;
+} my_match_t;
 
 enum my_lex_states
 {
@@ -116,9 +122,10 @@ typedef struct my_collation_handler_st
 
   int  (*strcasecmp)(struct charset_info_st *, const char *, const char *);
   
-  int  (*instr)(struct charset_info_st *,
+  uint (*instr)(struct charset_info_st *,
                 const char *big,   uint b_length,
-                const char *small, uint s_length);
+                const char *small, uint s_length,
+                my_match_t *match, uint nmatch);
   
   /* Hash calculation */
   void (*hash_sort)(struct charset_info_st *cs, const uchar *key, uint len,
@@ -249,9 +256,10 @@ extern void my_hash_sort_simple(CHARSET_INFO *cs,
 
 extern uint my_lengthsp_8bit(CHARSET_INFO *cs, const char *ptr, uint length);
 
-extern int  my_instr_simple(struct charset_info_st *,
+extern uint my_instr_simple(struct charset_info_st *,
                             const char *big,   uint b_length,
-                            const char *small, uint s_length);
+                            const char *small, uint s_length,
+                            my_match_t *match, uint nmatch);
 
 
 /* Functions for 8bit */
@@ -317,9 +325,10 @@ int my_wildcmp_mb(CHARSET_INFO *,
 		  int escape, int w_one, int w_many);
 uint my_numchars_mb(CHARSET_INFO *, const char *b, const char *e);
 uint my_charpos_mb(CHARSET_INFO *, const char *b, const char *e, uint pos);
-int  my_instr_mb(struct charset_info_st *,
+uint my_instr_mb(struct charset_info_st *,
                  const char *big,   uint b_length,
-                 const char *small, uint s_length);
+                 const char *small, uint s_length,
+                 my_match_t *match, uint nmatch);
 
 
 extern my_bool my_parse_charset_xml(const char *bug, uint len,

sql/item_func.cc

@@ -1161,7 +1161,7 @@ longlong Item_func_locate::val_int()
   null_value=0;
   uint start=0;
   uint start0=0;
-  int  ind;
+  my_match_t match;
 
   if (arg_count == 3)
   {
@@ -1175,11 +1175,12 @@ longlong Item_func_locate::val_int()
   if (!b->length())				// Found empty string at start
     return (longlong) (start+1);
   
-  ind= cmp_collation.collation->coll->instr(cmp_collation.collation,
+  if (!cmp_collation.collation->coll->instr(cmp_collation.collation,
                                             a->ptr()+start, a->length()-start,
-                                            b->ptr(), b->length());
-
-  return (longlong) (ind >= 0 ? ind + start0 + 1 : ind + 1);
+                                            b->ptr(), b->length(),
+                                            &match, 1))
+    return 0;
+  return (longlong) match.mblen + start0 + 1;
 }
 
 

strings/ctype-bin.c

@@ -263,16 +263,25 @@ static int my_strnxfrm_bin(CHARSET_INFO *cs __attribute__((unused)),
 }
 
 static
-int my_instr_bin(CHARSET_INFO *cs __attribute__((unused)),
+uint my_instr_bin(CHARSET_INFO *cs __attribute__((unused)),
                  const char *big,   uint b_length, 
-		 const char *small, uint s_length)
+		 const char *small, uint s_length,
+		 my_match_t *match, uint nmatch)
 {
   register const uchar *str, *search, *end, *search_end;
   
   if (s_length <= b_length)
   {
     if (!s_length)
-      return 0;		/* Empty string is always found */
+    {
+      if (nmatch)
+      {
+        match->beg= 0;
+        match->end= 0;
+        match->mblen= 0;
+      }
+      return 1;		/* Empty string is always found */
+    }
     
     str= (const uchar*) big;
     search= (const uchar*) small;
@@ -293,11 +302,24 @@ skipp:
 	  if ((*i++) != (*j++))
             goto skipp;
         
-	return (int) (str- (const uchar*)big) -1;
+        if (nmatch > 0)
+	{
+	  match[0].beg= 0;
+	  match[0].end= str- (const uchar*)big-1;
+	  match[0].mblen= match[0].end;
+	  
+	  if (nmatch > 1)
+	  {
+	    match[1].beg= match[0].end;
+	    match[1].end= match[0].end+s_length;
+	    match[1].mblen= match[1].end-match[1].beg;
+	  }
+	}
+	return 2;
       }
     }
   }
-  return -1;
+  return 0;
 }
 
 

strings/ctype-mb.c

@@ -274,18 +274,28 @@ uint my_charpos_mb(CHARSET_INFO *cs __attribute__((unused)),
   return b-b0;
 }
 
-int my_instr_mb(CHARSET_INFO *cs,
-                const char *big,   uint b_length, 
-                const char *small, uint s_length)
+uint my_instr_mb(CHARSET_INFO *cs,
+                 const char *big,   uint b_length, 
+                 const char *small, uint s_length,
+                 my_match_t *match, uint nmatch)
 {
-  register const char *end;
+  register const char *end, *big0;
   int res= 0;
   
   if (s_length <= b_length)
   {
     if (!s_length)
-      return 0;		// Empty string is always found
+    {
+      if (nmatch)
+      {
+        match->beg= 0;
+        match->end= 0;
+        match->mblen= 0;
+      }
+      return 1;		// Empty string is always found
+    }
     
+    big0= big;
     end= big+b_length-s_length+1;
     
     while (big < end)
@@ -294,15 +304,28 @@ int my_instr_mb(CHARSET_INFO *cs,
       
       if (!cs->coll->strnncoll(cs, (unsigned char*) big,   s_length, 
       				   (unsigned char*) small, s_length))
-        return res;
-      
+      {
+        if (nmatch)
+        {
+          match[0].beg= big0;
+          match[0].end= big-big0;
+          match[0].mblen= res;
+          if (nmatch > 1)
+          {
+            match[1].beg= match[0].end;
+            match[1].end= match[0].end+s_length;
+            match[1].mblen= 0;	/* Not computed */
+          }
+        }
+        return 2;
+      }
       mblen= (mblen= my_ismbchar(cs, big, end)) ? mblen : 1;
       big+= mblen;
       b_length-= mblen;
       res++;
     }
   }
-  return -1;
+  return 0;
 }
 
 /* BINARY collations handlers for MB charsets */

strings/ctype-simple.c

@@ -1030,16 +1030,25 @@ uint my_lengthsp_8bit(CHARSET_INFO *cs __attribute__((unused)),
 }
 
 
-int my_instr_simple(CHARSET_INFO *cs,
+uint my_instr_simple(CHARSET_INFO *cs,
                     const char *big,   uint b_length, 
-		    const char *small, uint s_length)
+		    const char *small, uint s_length,
+		    my_match_t *match, uint nmatch)
 {
   register const uchar *str, *search, *end, *search_end;
   
   if (s_length <= b_length)
   {
     if (!s_length)
-      return 0;		// Empty string is always found
+    {
+      if (nmatch)
+      {
+        match->beg= 0;
+        match->end= 0;
+        match->mblen= 0;
+      }
+      return 1;		/* Empty string is always found */
+    }
     
     str= (const uchar*) big;
     search= (const uchar*) small;
@@ -1060,11 +1069,24 @@ skipp:
 	  if (cs->sort_order[*i++] != cs->sort_order[*j++]) 
             goto skipp;
         
-	return (int) (str- (const uchar*)big) -1;
+	if (nmatch > 0)
+	{
+	  match[0].beg= 0;
+	  match[0].end= str- (const uchar*)big-1;
+	  match[0].mblen= match[0].end;
+	  
+	  if (nmatch > 1)
+	  {
+	    match[1].beg= match[0].end;
+	    match[1].end= match[0].end+s_length;
+	    match[1].mblen= match[1].end-match[1].beg;
+	  }
+	}
+	return 2;
       }
     }
   }
-  return -1;
+  return 0;
 }