stringcmp() and sortcmp() have been unified

into the only one sortcmp() with additional
CHARSET_INFO *cmp_charset argument.

sql/item.cc

@@ -120,8 +120,8 @@ bool Item_string::eq(const Item *item, bool binary_cmp) const
   if (type() == item->type())
   {
     if (binary_cmp)
-      return !stringcmp(&str_value, &item->str_value);
-    return !sortcmp(&str_value, &item->str_value);
+      return !sortcmp(&str_value, &item->str_value, &my_charset_bin);
+    return !sortcmp(&str_value, &item->str_value, charset());
   }
   return 0;
 }
@@ -1355,7 +1355,7 @@ bool field_is_equal_to_item(Field *field,Item *item)
     if (item->null_value)
       return 1;					// This must be true
     field->val_str(&field_tmp,&field_tmp);
-    return !stringcmp(&field_tmp,item_result);
+    return !sortcmp(&field_tmp,item_result,&my_charset_bin);
   }
   if (res_type == INT_RESULT)
     return 1;					// Both where of type int

sql/item_buff.cc

@@ -56,10 +56,8 @@ bool Item_str_buff::cmp(void)
   }
   else if (null_value)
     return 0;					// new and old value was null
-  else if (!item->binary())
-    tmp= sortcmp(&value,res) != 0;
   else
-    tmp= stringcmp(&value,res) != 0;
+    tmp= sortcmp(&value,res,item->charset()) != 0;
   if (tmp)
     value.copy(*res);				// Remember for next cmp
   return tmp;

sql/item_cmpfunc.cc

@@ -125,7 +125,9 @@ void Item_bool_func2::fix_length_and_dec()
     }
   }
   set_cmp_func();
-  binary_cmp= args[0]->binary() || args[1]->binary();
+  /* QQ: COERCIBILITY */
+  cmp_charset= (args[0]->binary() || args[1]->binary()) ? 
+	      &my_charset_bin : args[0]->charset();
 }
 
 
@@ -167,7 +169,7 @@ int Arg_comparator::compare_string()
     if ((res2= (*b)->val_str(&owner->tmp_value2)))
     {
       owner->null_value= 0;
-      return owner->binary_cmp ? stringcmp(res1,res2) : sortcmp(res1,res2);
+      return sortcmp(res1,res2,owner->cmp_charset);
     }
   }
   owner->null_value= 1;
@@ -181,8 +183,7 @@ int Arg_comparator::compare_e_string()
   res2= (*b)->val_str(&owner->tmp_value2);
   if (!res1 || !res2)
     return test(res1 == res2);
-  return (owner->binary_cmp ? test(stringcmp(res1, res2) == 0) :
-	    test(sortcmp(res1, res2) == 0));
+  return test(sortcmp(res1, res2, owner->cmp_charset) == 0);
 }
 
 
@@ -402,7 +403,7 @@ longlong Item_func_strcmp::val_int()
     null_value=1;
     return 0;
   }
-  int value= binary_cmp ? stringcmp(a,b) : sortcmp(a,b);
+  int value= sortcmp(a,b,cmp_charset);
   null_value=0;
   return !value ? 0 : (value < 0 ? (longlong) -1 : (longlong) 1);
 }
@@ -482,10 +483,11 @@ void Item_func_between::fix_length_and_dec()
   cmp_type=item_cmp_type(args[0]->result_type(),
            item_cmp_type(args[1]->result_type(),
                          args[2]->result_type()));
+  /* QQ: COERCIBILITY */
   if (args[0]->binary() | args[1]->binary() | args[2]->binary())
-    string_compare=stringcmp;
+    cmp_charset= &my_charset_bin;
   else
-    string_compare=sortcmp;
+    cmp_charset= args[0]->charset();
 
   /*
     Make a special case of compare with date/time and longlong fields.
@@ -517,17 +519,17 @@ longlong Item_func_between::val_int()
     a=args[1]->val_str(&value1);
     b=args[2]->val_str(&value2);
     if (!args[1]->null_value && !args[2]->null_value)
-      return (string_compare(value,a) >= 0 && string_compare(value,b) <= 0) ?
-	1 : 0;
+      return (sortcmp(value,a,cmp_charset) >= 0 && 
+	      sortcmp(value,b,cmp_charset) <= 0) ? 1 : 0;
     if (args[1]->null_value && args[2]->null_value)
       null_value=1;
     else if (args[1]->null_value)
     {
-      null_value= string_compare(value,b) <= 0; // not null if false range.
+      null_value= sortcmp(value,b,cmp_charset) <= 0; // not null if false range.
     }
     else
     {
-      null_value= string_compare(value,a) >= 0; // not null if false range.
+      null_value= sortcmp(value,a,cmp_charset) >= 0; // not null if false range.
     }
   }
   else if (cmp_type == INT_RESULT)
@@ -809,12 +811,13 @@ Item *Item_func_case::find_item(String *str)
       }
       if ((tmp=args[i]->val_str(str)))		// If not null
       {
+	/* QQ: COERCIBILITY */
 	if (first_expr_is_binary || args[i]->binary())
 	{
-	  if (stringcmp(tmp,first_expr_str)==0)
+	  if (sortcmp(tmp,first_expr_str,&my_charset_bin)==0)
 	    return args[i+1];
 	}
-	else if (sortcmp(tmp,first_expr_str)==0)
+	else if (sortcmp(tmp,first_expr_str,tmp->charset())==0)
 	  return args[i+1];
       }
       break;
@@ -1211,10 +1214,7 @@ cmp_item* cmp_item::get_comparator(Item *item)
 {
   switch (item->result_type()) {
   case STRING_RESULT:
-    if (item->binary())
-      return new cmp_item_binary_string;
-    else
-      return new cmp_item_sort_string;
+    return new cmp_item_sort_string(item->charset());
     break;
   case INT_RESULT:
     return new cmp_item_int;
@@ -1234,12 +1234,7 @@ cmp_item* cmp_item::get_comparator(Item *item)
 
 cmp_item* cmp_item_sort_string::make_same()
 {
-  return new cmp_item_sort_string_in_static();
-}
-
-cmp_item* cmp_item_binary_string::make_same()
-{
-  return new cmp_item_binary_string_in_static();
+  return new cmp_item_sort_string_in_static(cmp_charset);
 }
 
 cmp_item* cmp_item_int::make_same()
@@ -1346,6 +1341,22 @@ bool Item_func_in::nulls_in_row()
   return 0;
 }
 
+static int srtcmp_in(const String *x,const String *y)
+{
+  CHARSET_INFO *cs= x->charset();
+  return cs->strnncollsp(cs,
+                        (unsigned char *) x->ptr(),x->length(),
+			(unsigned char *) y->ptr(),y->length());
+}
+
+static int bincmp_in(const String *x,const String *y)
+{
+  CHARSET_INFO *cs= &my_charset_bin;
+  return cs->strnncollsp(cs,
+                        (unsigned char *) x->ptr(),x->length(),
+			(unsigned char *) y->ptr(),y->length());
+}
+
 void Item_func_in::fix_length_and_dec()
 {
   /*
@@ -1357,9 +1368,9 @@ void Item_func_in::fix_length_and_dec()
     switch (item->result_type()) {
     case STRING_RESULT:
       if (item->binary())
-	array=new in_string(arg_count,(qsort_cmp) stringcmp);
+	array=new in_string(arg_count,(qsort_cmp) srtcmp_in);
       else
-	array=new in_string(arg_count,(qsort_cmp) sortcmp);
+	array=new in_string(arg_count,(qsort_cmp) bincmp_in);
       break;
     case INT_RESULT:
       array= new in_longlong(arg_count);
@@ -1759,8 +1770,9 @@ bool Item_func_like::fix_fields(THD *thd, TABLE_LIST *tlist, Item ** ref)
 
   /*
     Comparision is by default done according to character set of LIKE
+    QQ: COERCIBILITY
   */
-  if (binary_cmp)
+  if (cmp_charset == &my_charset_bin)
     set_charset(&my_charset_bin);
   else
     set_charset(args[1]->charset());
@@ -1877,7 +1889,7 @@ longlong Item_func_regex::val_int()
       null_value=1;
       return 0;
     }
-    if (!regex_compiled || stringcmp(res2,&prev_regexp))
+    if (!regex_compiled || sortcmp(res2,&prev_regexp,&my_charset_bin))
     {
       prev_regexp.copy(*res2);
       if (regex_compiled)
@@ -1936,7 +1948,7 @@ void Item_func_like::turboBM_compute_suffixes(int *suff)
 
   *splm1 = pattern_len;
 
-  if (binary_cmp)
+  if (cmp_charset == &my_charset_bin)
   {
     int i;
     for (i = pattern_len - 2; i >= 0; i--)
@@ -2039,7 +2051,7 @@ void Item_func_like::turboBM_compute_bad_character_shifts()
   for (i = bmBc; i < end; i++)
     *i = pattern_len;
 
-  if (binary_cmp)
+  if (cmp_charset == &my_charset_bin)
   {
     for (j = 0; j < plm1; j++)
       bmBc[(uint) (uchar) pattern[j]] = plm1 - j;
@@ -2070,7 +2082,7 @@ bool Item_func_like::turboBM_matches(const char* text, int text_len) const
   const int tlmpl= text_len - pattern_len;
 
   /* Searching */
-  if (binary_cmp)
+  if (cmp_charset == &my_charset_bin)
   {
     while (j <= tlmpl)
     {

sql/item_cmpfunc.h

@@ -112,7 +112,7 @@ class Item_bool_func2 :public Item_int_func
 protected:
   Arg_comparator cmp;
   String tmp_value1,tmp_value2;
-  bool binary_cmp;
+  CHARSET_INFO *cmp_charset;
 
 public:
   Item_bool_func2(Item *a,Item *b):
@@ -127,7 +127,7 @@ public:
   bool have_rev_func() const { return rev_functype() != UNKNOWN_FUNC; }
   void print(String *str) { Item_func::print_op(str); }
   bool is_null() { return test(args[0]->is_null() || args[1]->is_null()); }
-  virtual bool binary() const { return binary_cmp; }
+  virtual bool binary() const { return test(cmp_charset->state & MY_CS_BINSORT); }
 
   static Item_bool_func2* eq_creator(Item *a, Item *b);
   static Item_bool_func2* ne_creator(Item *a, Item *b);
@@ -242,7 +242,7 @@ public:
 
 class Item_func_between :public Item_int_func
 {
-  int (*string_compare)(const String *x,const String *y);
+  CHARSET_INFO *cmp_charset;
 public:
   Item_result cmp_type;
   String value0,value1,value2;
@@ -263,7 +263,9 @@ public:
   void fix_length_and_dec()
   {
     max_length=2;
-    binary_cmp= args[0]->binary() || args[1]->binary();
+    /* QQ: COERCIBILITY */
+    cmp_charset= args[0]->binary() || args[1]->binary() ? 
+    		 &my_charset_bin : args[0]->charset();
   }
   optimize_type select_optimize() const { return OPTIMIZE_NONE; }
   const char *func_name() const { return "strcmp"; }
@@ -439,7 +441,8 @@ public:
 class cmp_item :public Sql_alloc
 {
 public:
-  cmp_item() {}
+  CHARSET_INFO *cmp_charset;
+  cmp_item() { cmp_charset= &my_charset_bin; }
   virtual ~cmp_item() {}
   virtual void store_value(Item *item)= 0;
   virtual int cmp(Item *item)= 0;
@@ -453,18 +456,14 @@ public:
   }
 };
 
-typedef int (*str_cmp_func_pointer)(const String *, const String *);
 class cmp_item_string :public cmp_item 
 {
 protected:
-  str_cmp_func_pointer str_cmp_func;
   String *value_res;
 public:
-  cmp_item_string (str_cmp_func_pointer cmp): str_cmp_func(cmp) {}
+  cmp_item_string (CHARSET_INFO *cs) { cmp_charset= cs; }
   friend class cmp_item_sort_string;
-  friend class cmp_item_binary_string;
   friend class cmp_item_sort_string_in_static;
-  friend class cmp_item_binary_string_in_static;
 };
 
 class cmp_item_sort_string :public cmp_item_string
@@ -473,12 +472,9 @@ protected:
   char value_buff[80];
   String value;
 public:
-  cmp_item_sort_string(str_cmp_func_pointer cmp):
-    cmp_item_string(cmp),
-    value(value_buff, sizeof(value_buff), default_charset_info) {}
-  cmp_item_sort_string():
-    cmp_item_string(&sortcmp),
-    value(value_buff, sizeof(value_buff), default_charset_info) {}
+  cmp_item_sort_string(CHARSET_INFO *cs):
+    cmp_item_string(cs),
+    value(value_buff, sizeof(value_buff), cs) {}
   void store_value(Item *item)
   {
     value_res= item->val_str(&value);
@@ -489,22 +485,16 @@ public:
     String tmp(buff, sizeof(buff), default_charset_info), *res;
     if (!(res= arg->val_str(&tmp)))
       return 1;				/* Can't be right */
-    return (*str_cmp_func)(value_res, res);
+    return sortcmp(value_res, res, cmp_charset);
   }
   int compare(cmp_item *c)
   {
     cmp_item_string *cmp= (cmp_item_string *)c;
-    return (*str_cmp_func)(value_res, cmp->value_res);
+    return sortcmp(value_res, cmp->value_res, cmp_charset);
   } 
   cmp_item *make_same();
 };
 
-class cmp_item_binary_string :public cmp_item_sort_string {
-public:
-  cmp_item_binary_string(): cmp_item_sort_string(&stringcmp)  {}
-  cmp_item *make_same();
-};
-
 class cmp_item_int :public cmp_item
 {
   longlong value;
@@ -590,9 +580,8 @@ class cmp_item_sort_string_in_static :public cmp_item_string
  protected:
   String value;
 public:
-  cmp_item_sort_string_in_static(str_cmp_func_pointer cmp):
-    cmp_item_string(cmp) {}
-  cmp_item_sort_string_in_static(): cmp_item_string(&sortcmp) {}
+  cmp_item_sort_string_in_static(CHARSET_INFO *cs):
+    cmp_item_string(cs) {}
   void store_value(Item *item)
   {
     value_res= item->val_str(&value);
@@ -606,21 +595,11 @@ public:
   int compare(cmp_item *c)
   {
     cmp_item_string *cmp= (cmp_item_string *)c;
-    return (*str_cmp_func)(value_res, cmp->value_res);
+    return sortcmp(value_res, cmp->value_res, cmp_charset);
   }
   cmp_item * make_same()
   {
-    return new cmp_item_sort_string_in_static();
-  }
-};
-
-class cmp_item_binary_string_in_static :public cmp_item_sort_string_in_static {
-public:
-  cmp_item_binary_string_in_static():
-    cmp_item_sort_string_in_static(&stringcmp) {}
-  cmp_item * make_same()
-  {
-    return new cmp_item_binary_string_in_static();
+    return new cmp_item_sort_string_in_static(cmp_charset);
   }
 };
 

sql/item_func.cc

@@ -107,7 +107,7 @@ Item_func::fix_fields(THD *thd, TABLE_LIST *tables, Item **ref)
     return 0;					// Fatal error if flag is set!
   if (arg_count)
   {						// Print purify happy
-    coercibility= COER_NOCOLL;
+    bool first_coll= 1;
     for (arg=args, arg_end=args+arg_count; arg != arg_end ; arg++)
     {
       if ((*arg)->fix_fields(thd, tables, arg) ||
@@ -122,15 +122,17 @@ Item_func::fix_fields(THD *thd, TABLE_LIST *tables, Item **ref)
 	  Set return character set to first argument if we are returning a
 	  string.
 	*/
-	if ((*arg)->binary())
+	if (first_coll)
 	{
-	  set_charset(&my_charset_bin);
-	  coercibility= COER_NOCOLL;
+	  set_charset((*arg)->charset());
+	  coercibility= (*args)->coercibility;
+	  first_coll= 0;
 	}
-	else if (coercibility== COER_NOCOLL)
+	else if ((*arg)->charset() == &my_charset_bin || 
+		 charset() == &my_charset_bin)
 	{
-	  coercibility= (*arg)->coercibility;
-	  set_charset((*arg)->charset());
+	  set_charset(&my_charset_bin);
+	  coercibility= COER_NOCOLL;
 	}
 	else if ((*arg)->coercibility < coercibility)
 	{
@@ -864,6 +866,7 @@ void Item_func_min_max::fix_length_and_dec()
   max_length=0;
   maybe_null=1;
   cmp_type=args[0]->result_type();
+
   for (uint i=0 ; i < arg_count ; i++)
   {
     if (max_length < args[i]->max_length)
@@ -873,11 +876,11 @@ void Item_func_min_max::fix_length_and_dec()
     if (!args[i]->maybe_null)
       maybe_null=0;
     cmp_type=item_cmp_type(cmp_type,args[i]->result_type());
-    if (args[i]->binary())
+    if (i==0)
+      set_charset(args[i]->charset());
+    else if (args[i]->charset() == &my_charset_bin)
       set_charset(&my_charset_bin);
   }
-  if (cmp_type == STRING_RESULT)
-    str_cmp_function= binary() ? stringcmp : sortcmp;
 }
 
 
@@ -922,7 +925,7 @@ String *Item_func_min_max::val_str(String *str)
 	res2= args[i]->val_str(res == str ? &tmp_value : str);
 	if (res2)
 	{
-	  int cmp= (*str_cmp_function)(res,res2);
+	  int cmp= sortcmp(res,res2,charset());
 	  if ((cmp_sign < 0 ? cmp : -cmp) < 0)
 	    res=res2;
 	}

sql/item_func.h

@@ -524,7 +524,6 @@ class Item_func_min_max :public Item_func
   Item_result cmp_type;
   String tmp_value;
   int cmp_sign;
-  int (*str_cmp_function)(const String *x,const String *y);
 public:
   Item_func_min_max(List<Item> &list,int cmp_sign_arg) :Item_func(list),
     cmp_type(INT_RESULT), cmp_sign(cmp_sign_arg) {}

sql/item_sum.cc

@@ -183,12 +183,17 @@ Item_sum_hybrid::fix_fields(THD *thd, TABLE_LIST *tables, Item **ref)
     return 1;
   hybrid_type=item->result_type();
   if (hybrid_type == INT_RESULT)
+  {
+    cmp_charset= &my_charset_bin;
     max_length=20;
+  }
   else if (hybrid_type == REAL_RESULT)
+  {
+    cmp_charset= &my_charset_bin;
     max_length=float_length(decimals);
-  else
+  }else
   {
-    str_cmp_function= item->binary() ? stringcmp : sortcmp;
+    cmp_charset= item->charset();
     max_length=item->max_length;
   }
   decimals=item->decimals;
@@ -440,7 +445,7 @@ bool Item_sum_min::add()
   {
     String *result=args[0]->val_str(&tmp_value);
     if (!args[0]->null_value &&
-	(null_value || (*str_cmp_function)(&value,result) > 0))
+	(null_value || sortcmp(&value,result,cmp_charset) > 0))
     {
       value.copy(*result);
       null_value=0;
@@ -487,7 +492,7 @@ bool Item_sum_max::add()
   {
     String *result=args[0]->val_str(&tmp_value);
     if (!args[0]->null_value &&
-	(null_value || (*str_cmp_function)(&value,result) < 0))
+	(null_value || sortcmp(&value,result,cmp_charset) < 0))
     {
       value.copy(*result);
       null_value=0;
@@ -762,7 +767,7 @@ Item_sum_hybrid::min_max_update_str_field(int offset)
     result_field->ptr-=offset;
 
     if (result_field->is_null() ||
-	(cmp_sign * (*str_cmp_function)(res_str,&tmp_value)) < 0)
+	(cmp_sign * sortcmp(res_str,&tmp_value,cmp_charset)) < 0)
       result_field->store(res_str->ptr(),res_str->length(),res_str->charset());
     else
     {						// Use old value

sql/item_sum.h

@@ -369,17 +369,19 @@ class Item_sum_hybrid :public Item_sum
   enum_field_types hybrid_field_type;
   int cmp_sign;
   table_map used_table_cache;
-  int (*str_cmp_function)(const String *x,const String *y);
+  CHARSET_INFO *cmp_charset;
 
   public:
   Item_sum_hybrid(Item *item_par,int sign)
     :Item_sum(item_par), hybrid_type(INT_RESULT), cmp_sign(sign),
-    used_table_cache(~(table_map) 0)
+    used_table_cache(~(table_map) 0),
+    cmp_charset(&my_charset_bin)
   {}
   Item_sum_hybrid(THD *thd, Item_sum_hybrid &item):
     Item_sum(thd, item), value(item.value), tmp_value(item.tmp_value),
     sum(item.sum), sum_int(item.sum_int), hybrid_type(item.hybrid_type),
-    cmp_sign(item.cmp_sign), used_table_cache(used_table_cache) {}
+    cmp_sign(item.cmp_sign), used_table_cache(used_table_cache),
+    cmp_charset(item.cmp_charset) {}
   bool fix_fields(THD *, TABLE_LIST *, Item **);
   table_map used_tables() const { return used_table_cache; }
   bool const_item() const { return !used_table_cache; }

sql/sql_analyse.cc

@@ -41,13 +41,13 @@
 int sortcmp2(void* cmp_arg __attribute__((unused)),
 	     const String *a,const String *b)
 {
-  return sortcmp(a,b);
+  return sortcmp(a,b,a->charset());
 }
 
 int stringcmp2(void* cmp_arg __attribute__((unused)),
 	     const String *a,const String *b)
 {
-  return stringcmp(a,b);
+  return sortcmp(a,b,&my_charset_bin);
 }
 
 int compare_double2(void* cmp_arg __attribute__((unused)),
@@ -329,20 +329,10 @@ void field_str::add()
     if (length > max_length)
       max_length = length;
 
-    if (item->binary())
-    {
-      if (stringcmp(res, &min_arg) < 0)
-	min_arg.copy(*res);
-      if (stringcmp(res, &max_arg) > 0)
-	max_arg.copy(*res);
-    }
-    else
-    {
-      if (sortcmp(res, &min_arg) < 0)
-	min_arg.copy(*res);
-      if (sortcmp(res, &max_arg) > 0)
-	max_arg.copy(*res);
-    }
+    if (sortcmp(res, &min_arg,item->charset()) < 0)
+      min_arg.copy(*res);
+    if (sortcmp(res, &max_arg,item->charset()) > 0)
+      max_arg.copy(*res);
   }
 
   if (room_in_tree)

sql/sql_string.cc

@@ -538,30 +538,14 @@ void String::qs_append(const char &c)
 }
 
 
-int sortcmp(const String *x,const String *y)
+int sortcmp(const String *x,const String *y, CHARSET_INFO *cs)
 {
-  CHARSET_INFO *cs= x->str_charset;
   return cs->strnncollsp(cs,
                         (unsigned char *) x->ptr(),x->length(),
 			(unsigned char *) y->ptr(),y->length());
 }
 
 
-int stringcmp(const String *x,const String *y)
-{
-  const char *s= x->ptr();
-  const char *t= y->ptr();
-  uint32 x_len=x->length(),y_len=y->length(),len=min(x_len,y_len);
-
-  while (len--)
-  {
-    if (*s++ != *t++)
-      return ((int) (uchar) s[-1] - (int) (uchar) t[-1]);
-  }
-  return (int) (x_len-y_len);
-}
-
-
 String *copy_if_not_alloced(String *to,String *from,uint32 from_length)
 {
   if (from->Alloced_length >= from_length)

sql/sql_string.h

@@ -25,8 +25,7 @@
 #endif
 
 class String;
-int sortcmp(const String *a,const String *b);
-int stringcmp(const String *a,const String *b);
+int sortcmp(const String *a,const String *b, CHARSET_INFO *cs);
 String *copy_if_not_alloced(String *a,String *b,uint32 arg_length);
 
 class String
@@ -205,8 +204,7 @@ public:
   void strip_sp();
   inline void caseup() { my_caseup(str_charset,Ptr,str_length); }
   inline void casedn() { my_casedn(str_charset,Ptr,str_length); }
-  friend int sortcmp(const String *a,const String *b);
-  friend int stringcmp(const String *a,const String *b);
+  friend int sortcmp(const String *a,const String *b, CHARSET_INFO *cs);
   friend String *copy_if_not_alloced(String *a,String *b,uint32 arg_length);
   uint32 numchars();
   int charpos(int i,uint32 offset=0);